Heart Rate Zones

A heart rate zone is a band of heart rates that stands in for a level of effort. The trouble is that there is no single set of zones. Researchers usually use three, most watches and plans use five, and some coaches use six or seven, and each system puts its boundaries in different places. "Zone 2" in a training plan and "Zone 2" on your watch can be different efforts.

Only two boundaries have a physiological basis. The first lactate threshold (LT1) is roughly where breathing and blood lactate first rise above their easy-running levels, and the second (LT2) is roughly the hardest effort you can hold steadily for a long time (see Lactate Threshold). Everything else, including where Zone 1 becomes Zone 2 on a five-zone watch, is a convention chosen by a coach, a federation or a manufacturer. That is why one easy run can honestly be labelled Zone 1, Zone 2 or Zone 3.

If you have not tested your maximum heart rate, do not train to the default zones your watch or app calculates. They are percentages of a maximum that is usually estimated from your age, and they can be misleading for you (see How much to trust your settings). If you have not had an LT1 or LT2 test, use the talk test rather than calculated zones, both to keep easy runs easy and to set your own zones.

What to do now. Before you follow any zone instruction, check three things: how the plan defines its zones, which sport profile and basis your watch is using (percentage of maximum heart rate, heart rate reserve or threshold heart rate), and the actual beats-per-minute boundaries on the watch. To keep things clear, Fellrnr.com writes zones as "Zone 2 of 5" (or "Z2 of 5"), adding the system where the count alone is ambiguous: "Zone 2 of 5 (Polar, 60–70% of maximum heart rate)" or "Zone 2 of 3 (between LT1 and LT2)". Where an instruction simply means easy running, it says "below LT1". Fellrnr's own five-zone scale, built on your thresholds, names its zones Recovery, Easy, Aerobic, Threshold and Maximum (see my recommendation).

A personal view. As the statistician George Box put it, "all models are wrong, but some are useful". I have found heart rate extremely useful in my own training, but only as a model of effort and only as one piece of information among several. Your heart rate and your zones are there to serve you as a runner and to guide your training, not to control it. Understanding how your heart rate relates to your effort, and how it changes with fatigue, dehydration, training and recovery, matters more than the science. Even measured thresholds change from day to day and from month to month, so always pay attention to what your breathing is telling you.

Disclosure: this article was created using iterative research loops using AI, with AI writing up the final conclusions. ChatGPT and Claude were used.

One runner, three labels. At a settled 140 bpm this runner is in Z1 of 3 (below LT1), Z2 of 5 on a custom scale built on their thresholds, and Z3 of 5 on Polar's default (70–80% of maximum heart rate). Colours show the zone number, using paler versions of Garmin's zone colours. Bold boundaries are measured for this runner, starred ones are illustrative, and the Polar boundaries come from its published percentages. The same data are in the table below.

How to use zones on your next run

Start with the purpose of the run

A zone is a means of hitting the effort the workout is meant to have. Work out the purpose first, then pick the tool.

Your situation	What to do
You are following a named plan	Find the plan's zone definitions (its glossary, the book or the coach). If it only says "Zone 2", ask the author. Set your watch to the plan's boundaries, not the factory ones.
You are using factory watch settings	Do not train to the factory zones. They are percentages of a maximum heart rate that is usually estimated from your age, so both the percentage and the maximum can be wrong for you, and the zones can mislead. Use the talk test for easy runs, and to set your own zones.
You have a credible running maximum but no thresholds	A measured maximum removes one error, but LT1 still sits at very different percentages in different runners, so calculated zones can still put easy running in the wrong place. Set your zones with the talk test and use percentages of maximum only as a rough check.
You have one or both thresholds	Use them. "Below LT1" and "between LT1 and LT2" are the instructions most plans are really giving. Set custom zones in bpm around your thresholds (see my recommendation).

Without test results, use the talk test

If you have not measured your thresholds, and especially if your maximum heart rate is an estimate, use the talk test rather than a calculated heart rate zone to keep easy runs easy. It needs no numbers, it responds to heat, hills and fatigue as they happen, and it tracks the first threshold, which is the boundary an easy run has to stay below. You can also use it to set your heart rate zones (see below).

How to do it:

1. Choose a passage of about 30 words that you know by heart, such as a nursery rhyme or a few lines of a poem, and use the same one each time.
2. Every few minutes, and after hills, say it aloud while running.
3. Ask yourself one question: can I speak comfortably? The answer is yes, not sure, or no.
4. For an easy run, stay where the answer is a clear yes. "Not sure" marks roughly the first threshold, and "no" means you are above it.

The evidence.

• In 16 healthy adults tested on both a treadmill and a cycle, "At the point where speech first became difficult, exercise intensity was almost exactly equivalent to ventilatory threshold", and when speech was not comfortable the intensity was consistently above it.
• In 14 well-trained runners, 40-minute runs at the last speed where speech was comfortable, or one stage slower, gave heart rates and perceived effort within accepted training ranges; at the "not sure" stage no runner could talk comfortably. Repeated tests agreed reasonably well (r = 0.71 for the talk test, 0.84 for heart rate).
• When running speed was set by the talk test alone, all 16 active adults completed 30 minutes at the last comfortable stage, 13 at the "not sure" stage and 2 at the uncomfortable stage.
• In 15 adults, the talk test matched the lactate threshold more closely than the ventilatory threshold, and the last comfortable stage averaged 82% of maximum heart rate. That is Z4 of 5 on Garmin and Polar defaults (see Watch defaults), for an effort these people could still talk through.
• In a 10-week training study of 44 young adults, cycling at the highest power that still allowed comfortable speech improved V̇O₂max about as much as training to percentages of heart rate reserve set from a measured maximum (about 11% in both groups).
• Heart rate zones lose accuracy when the maximum is estimated. In 165 recreational runners, every fixed anchor became less accurate when the maximum was estimated rather than measured, and the authors noted that perceived effort then "may perform better than HR- or speed-derived methods". Age equations have a typical error of about 11–12 bpm.

What this adds up to. The talk test locates the region of the first threshold reasonably well without any measured numbers: moderate evidence, from small studies, mostly in young, healthy adults. An age-estimated heart rate zone is typically wrong by 11–12 bpm: strong evidence. No study I found compared the two head to head in runners, so the conclusion that the talk test is the better guide for untested runners is an inference, though a well-supported one.

Its limits:

• It is best at marking the top of easy running. Its marker for the second threshold is rougher (see below), so use pace or perceived effort for harder sessions.
• "Not sure" is a band, not a line, and the answer depends on using the same passage each time.
• Much of the research comes from one group at the University of Wisconsin–La Crosse. The American Council on Exercise, a fitness certification organisation, describes that group's talk test work as ACE-sponsored research, without saying which studies it funded.

If you cannot keep running while the answer is a clear yes, alternate running and walking (see Walking Breaks); for some runners all continuous running is above easy effort (see Pace Range Ratio).

Using the talk test to set your heart rate zones

You can use the talk test to find the heart rates to put into your watch, so that your zones come from your own response rather than from a formula.

1. Use a treadmill or a flat route, and warm up for at least 10 minutes.
2. Run in stages of about 3 minutes, each a little faster than the last (for example, 0.5 km/h faster), so that heart rate has time to settle.
3. In the last 30 seconds of each stage, say your passage, answer the question (yes, not sure or no) and note your heart rate.
4. Stop at the first clear "no".
5. Repeat on another day in similar conditions, and average the two results.

Reading the result on the Fellrnr five-zone scale:

• The heart rate at your first "not sure" stage is your estimate of LT1. Use it as the boundary between Z2 (Easy) and Z3 (Aerobic), and keep easy runs at heart rates where the answer was a clear yes.
• The heart rate at your first "no" stage is a rough estimate of LT2. Use it, cautiously, as the boundary between Z3 (Aerobic) and Z4 (Threshold), and check it against your pace in races or time trials when you can.
• The other two cuts are coaching choices. The worked example puts the Recovery–Easy cut at about 90% of the LT1 heart rate. The Threshold–Maximum cut needs a credible maximum (it sits roughly halfway between LT2 and maximum); without one, treat everything above your LT2 estimate as hard and judge it by pace or effort.

The evidence. In 18 highly trained cyclists, workload, heart rate, lactate and perceived effort did not differ significantly between the first ventilatory threshold and the "equivocal" (not sure) stage, or between the respiratory compensation threshold, the ventilatory marker of the second threshold, and the "negative" stage. Heart rates correlated at r = 0.79 and 0.92, and the authors concluded that "the EQ and NEG stages of TT can be used as a simple and practical surrogate of the VT and RCT in highly trained cyclists". These were cyclists, not runners; no significant difference across a group does not guarantee a close match for each person; and Foster, a co-author, is from the La Crosse group. Together with the studies above that place the first threshold where speech first becomes difficult, this is moderate evidence for the LT1 estimate and limited evidence for the LT2 estimate in runners.

My opinion. If you have not had a lab test, I think heart rates taken from a talk test like this are a better basis for your zones than heart rates calculated from a formula, such as fixed percentages of an age-estimated maximum. The talk test measures your own response, while a formula assumes you are average, and age equations alone are typically wrong by 11–12 bpm. I found no study that compared the two head to head, so this is my judgement, built on the evidence above, not a tested finding. Check the result against how your runs feel over the following weeks, and repeat the test after a block of training or a long break.

A worked example

Take a hypothetical runner with a running maximum heart rate of 180 bpm, a resting heart rate of 50, LT1 at 145 bpm and LT2 at 165 bpm. On an easy run their heart rate settles at 140 bpm.

• Three zones: 140 is below LT1, so it is Z1 of 3.
• Polar or Garmin defaults: 140 ÷ 180 = 77.8% of maximum heart rate. Both defaults put 70–80% in Zone 3, so the watch shows Z3 of 5.
• A custom five-zone scale with cuts at 130, 145, 165 and 173 bpm: 140 is Z2 of 5, which the Fellrnr scale calls "Easy". The 145 and 165 cuts are this runner's thresholds. The 130 and 173 cuts are coaching choices (see below).
• Heart Rate Reserve: (140 − 50) ÷ (180 − 50) = 69.2% of reserve. Which zone that is depends entirely on the cut-offs of the reserve-based system in use, so it cannot be named without them.

The run is the same genuinely easy run in every case. The three targets you will meet are calculated like this:

Basis	Formula	Example for this runner
Percentage of maximum heart rate	fraction × maximum	70% → 0.70 × 180 = 126 bpm
Percentage of heart rate reserve	resting + fraction × (maximum − resting)	70% → 50 + 0.70 × 130 = 141 bpm
Percentage of threshold heart rate (LTHR)	fraction × LTHR	85% of 165 → 140 bpm

"70%" is 126 bpm on one basis and 141 on another. LTHR needs checking too: in most coaching systems it means the heart rate at the second threshold, often estimated from a 20- or 30-minute time trial, but a plan may define it differently.

In this article, a heart rate exactly on a boundary belongs to the higher zone, and bpm values are rounded to whole beats.

How much to trust your settings

Setting	Confidence	Why
Thresholds from a recent running test (laboratory or well-run field test)	Highest	Anchors zones to your own physiology, though thresholds vary with the test method and from day to day
Maximum heart rate reached in a hard running test or race, seen on a clean trace	Good for the maximum; moderate for zones	You know the maximum, but thresholds sit at different percentages of it in different runners
Highest value your watch has recorded	Uncertain	It is an observed peak. It may be below your true maximum, or an optical-sensor spike above it
Maximum estimated from age	Low for an individual	Right on average, often wrong for a person

• Maximum heart rate is sport-specific. Use a running value for running zones; Olympiatoppen notes an athlete can have a maximum about 5% higher running than cycling. See Maximum Heart Rate for testing guidance. A maximal test is hard and is not something every reader needs to do; if you have any doubt about your health, talk to a doctor first.
• A peak is not an established maximum. The highest number in your history only becomes your maximum when it comes from an all-out effort and the trace is believable. Wrist sensors can lock on to cadence and report a false high value (see Optical Heart Rate Monitoring and Bad heart rate data), so check a suspicious peak against a chest strap or a trace that rises and falls smoothly, and do not let your watch set your maximum automatically.
• Age equations. In 762 sedentary adults tested on a cycle ergometer, the equation 208 − 0.7 × age was almost exactly right on average (a difference of 0.03 bpm), but the standard error of estimate was 11.4 bpm, and 12.4 bpm for 220 − age. Those are typical errors, not the largest ones. An age equation gives you a starting guess when you have nothing better.
• What an error costs. If your assumed maximum is wrong by 12 bpm, an "80% of maximum" boundary moves by 0.80 × 12 = 9.6 bpm, before any allowance for where your own thresholds sit.
• You do not need a maximum to use thresholds. A threshold-based plan works from threshold heart rates directly. You only need a measured maximum to know your maximum.

When heart rate helps and when it misleads

What the tests estimate. A laboratory lactate test finds the points where blood lactate first rises (LT1) and then rises steeply (LT2) as speed increases; a gas-exchange test finds the matching changes in breathing, the first and second ventilatory thresholds (VT1 and VT2). They are related markers, not identical ones, and the answer depends on the protocol. The talk test approximates the same two points: speech becomes doubtful at about VT1 and comfortable speech is no longer possible at about VT2. A single field test, such as a 30-minute time trial, estimates one upper threshold. It tells you little about LT1, which cannot be reliably calculated from the upper threshold or from critical speed.

Run	Use heart rate?	Why
Easy, steady run in cool conditions	Yes.	Heart rate settles after a few minutes and tracks effort well
Long, hot or hilly sustained run	Check it against feel.	Heart rate drifts upwards over time and in heat at the same pace (see Heart Rate Drift); hills change it quickly
Short repetitions (under about 4 minutes)	No. Use pace, power or duration.	Heart rate lags at the start of each effort and is still rising when the repetition ends

Olympiatoppen describes heart rate as a good measure in continuous work or in intervals long enough for it to stabilise (more than 4 minutes), and notes that day-to-day variation can be 6–12 beats.

Minutes in a zone are not the purpose of an interval session. In 15 × 1 minute hard with 1-minute recoveries, a heart rate monitor might record only 7–8 minutes in the top zone, yet the session is 15 minutes of top-zone work. Judge an interval session by the work done, not by the time-in-zone chart.

Bad heart rate data

Every heart rate monitor sometimes records readings that are simply wrong, and a few bad minutes can distort your zones, your training load and your estimate of maximum heart rate.

Optical sensors on the wrist or arm shine light into the skin and detect the pulse in the blood vessels. Their main problems are:

• Locking on to cadence. Optical sensors "tend to lock on to the periodic signal stemming from the repetitive motion", so during a run they may report your step rate, often 160–190 per minute, as your heart rate.
• Movement in general. In one study, error during activity was on average 30% higher than at rest, while there was "no statistically significant difference in accuracy across skin tones".
• Fit and cold. Garmin advises wearing the watch above the wrist bone, "snug but comfortable", and warming up indoors in cold conditions before starting.

Chest straps measure the heart's electrical signal (an ECG) and are generally more accurate. In 50 healthy adults, a chest strap agreed with a medical ECG almost perfectly (r_c = 0.996) and better than any of the wrist-worn devices tested, none of which "was accurate during elliptical trainer use with arm levers". The authors concluded that "Electrode-containing chest monitors should be used when accurate HR measurement is imperative". In a study partly funded by Polar, a Polar H10 strap recorded clean beat-to-beat data 99.6% of the time, against 94.6% for a medical Holter ECG. Straps fail in their own ways. Polar lists a dirty or worn strap, poor contact between the skin and the electrodes (it advises dampening them before training), static from a fluttering shirt, a weak connection to the watch, and a low battery: "If the battery is running low, the transmission range decreases which may cause errors in heart rate readings".

An arm-worn optical sensor can come close. In a validation study, a Polar Verity Sense worn on the upper arm differed from a Polar H10 chest strap by an average of 1.35% (1.43 bpm) across activities from sitting to vigorous exercise, with the smallest errors while jogging; worn on the forearm, its agreement fell during high-intensity intervals. The study declares no conflicts of interest, but an earlier strap study by one of its authors was partly funded by Polar.

How bad data distorts training models. Any method that uses heart rate, from time in zones to the training load scores on watches and apps and the Training Budget, counts a bad reading as real effort. The damage is greatest when a reading is very high for a long time. For the worked runner (maximum 180), a 40-minute easy run at 130 bpm counts about 31 easy-running minutes on the Training Budget; if cadence lock records the same run at 175 bpm, it counts about 97, more than three times as much. A single high spike matters too, if your watch treats it as a new maximum.

Turn off automatic maximum heart rate detection. Garmin's watches, for example, can "automatically detect your maximum heart rate and lactate threshold during an activity", and "The Auto Detection feature is turned on by default". In my experience, automatic detection tends to pick up bad data, and a falsely high maximum moves every percentage-based zone upwards, so easy runs look easier than they are. I recommend turning it off and setting your maximum and thresholds yourself.

Check the trace. Signs of bad data include a jump of 50 bpm or more in a few seconds, a heart rate that matches your cadence, a value that stays flat for minutes, and a heart rate that does not fall when you stop. Remove or correct those sections, if your software allows it, before they feed your training load (see Fixing Heart Rate Monitor Problems and Data Quality).

My experience. I find ECG chest straps much better than any optical system I have used, although the Polar Verity Sense is good enough for most work. I have also found that a low battery in a chest strap seems to increase the error rate, which fits Polar's own guidance above.

Perceived effort and the talk test

Effort	Speech and breathing	Borg 6–20 scale	Borg CR10 scale (0–10)
Well below LT1	Can speak effortlessly	Below 11 ("light" or easier)	1–2
Just below LT1	Long sentences, relatively effortlessly	Below 13	2–3
At about LT1	Speech becomes doubtful	Mean 10.8 (SD 1.8) in one large study	Qualitative only
Between LT1 and LT2	Short sentences	13–14 ("somewhat hard")	4–5
At about LT2	Comfortable speech no longer possible	Mean 13.6 (SD 1.8) in the same study	Qualitative only
Above LT2	A few words, then one or two	14–19	6–10

• Sources. The ranges are Olympiatoppen's guidance for its intensity scale. The means at the thresholds come from 2,560 people tested on a treadmill or cycle. The speech stages come from a review of the talk test.
• The spread is wide. In 165 recreational runners, the 95% range of ratings was 9.4–15.7 at LT1 and 11.0–19.8 at LT2, and perceived effort was the least accurate of the fixed anchors tested. Use the numbers as typical values, not targets.
• Three different scales. The Borg 6–20 scale and the Borg CR10 scale have fixed verbal anchors. The informal "effort out of 10" used in many plans and apps has no standard anchors. Do not convert one scale into another by arithmetic; a CR10 rating is not a 6–20 rating divided by two.
• The "times ten" rule. The 6–20 scale was built so that a rating multiplied by ten roughly matched heart rate, from about 60 to 200 bpm, in healthy adults during steady exercise. A maximum of 200 fits some young adults. It cannot predict an individual's heart rate: a masters runner with a maximum of 160 reaches a rating of 19 or 20 at 160 bpm.
• During the run or afterwards? The table is about effort during the run. A rating for the whole session, given afterwards (session RPE), is a different measure used for training load (see Rating of Perceived Exertion).

What the different systems measure

Summary of systems

System	Zones and anchor	Boundaries	What "Z2" means	Main use	Main limitation
Research three-zone model	3; measured LT1 and LT2 (or VT1 and VT2)	Below the first threshold, between the two, above the second	Between LT1 and LT2: moderately hard	Describing training intensity distribution	Needs testing; thresholds depend on the method
Olympiatoppen I-scale, version 2 (2024)	8 (five aerobic, I-1 to I-5, plus three anaerobic); guidance values of % maximum heart rate, lactate, perceived effort and breathing	I-1 about 55–72%, I-2 72–82%, I-3 82–87%, I-4 87–92%, I-5 92–100% of maximum	About 72–82% of maximum, lactate about 1.0–2.0 mmol/L: steady, "fairly light"	Norwegian elite endurance sport; much research	Percentages come from elite athletes; meant to be individualised
Garmin default	5; % of maximum heart rate	50–60, 60–70, 70–80, 80–90, 90–100%	60–70% of maximum: for most runners very easy	Out-of-the-box display	Maximum usually estimated; bands unrelated to your thresholds
Polar default	5; % of maximum heart rate	50–60, 60–70, 70–80, 80–90, 90–100%	60–70% of maximum ("light")	Out-of-the-box display	As Garmin; nothing counted below 50%
Fellrnr five-zone scale (this article)	5 (Recovery, Easy, Aerobic, Threshold, Maximum); your own LT1 and LT2 plus two chosen cuts	Z1–2 below LT1, Z3 between LT1 and LT2, Z4–5 above LT2	"Easy": the upper part of the range below LT1	One scale that translates to the three-zone model	Needs thresholds, measured or estimated with the talk test; two cuts are arbitrary
Coggan power levels (cycling)	7; % of functional threshold power	See Power zones	"Endurance", 56–75% of threshold power	Cycling with a power meter	Power, not heart rate; does not transfer to running watts

The research three-zone model

The three-zone model divides exercise at two measured points: below the first threshold, between the thresholds, and above the second. Heart rate becomes a field proxy when you note the heart rate at each measured threshold and use those two numbers outdoors.

• The thresholds are operational definitions. There are many ways to locate a lactate threshold, and they give different values. LT2, VT2, the maximal lactate steady state and critical speed are related but not interchangeable (see Lactate Threshold and Critical Power). Fixed concentrations such as 2 and 4 mmol/L do not identify everyone's thresholds; typical values are about 1.0–2.0 mmol/L at the first and 2.5–4.0 at the second in trained athletes.
• Training zones are not quite the metabolic domains. Physiologists also describe moderate, heavy and severe domains. A critical review found good evidence only for critical power or speed as the boundary between heavy and severe. A three-zone training diagram and a domain diagram look alike, but each depends on how its boundaries were measured.
• Fixed percentages miss individuals. In 165 recreational runners, LT1 averaged about 80% and LT2 about 90% of maximum heart rate, but at exactly 80% of maximum, 57% of runners were below LT1 and 43% above it. Fixed anchors were wrong by 4.9–7.4 bpm on average at LT1 and 2.8–5.2 bpm at LT2, depending on the method, and by more when maximum heart rate was estimated. In 1,411 habitual runners, the heart rate at VT1 was 85.1 ± 4.6% of peak heart rate and at VT2 93.5 ± 2.5%. An individual error can be much larger than an average one. Strong evidence that no fixed percentage suits everyone.

Five-zone schemes and the Norwegian scale

In a survey of 710 endurance athletes, coaches and scientists, a five-zone scale was the most common (47%), followed by a three-zone scale. The sample was recruited through social media, was 80% male and over-represented Norway, so it describes these respondents, not all runners.

The best documented five-zone scheme is Norwegian. Seiler and Tønnessen's 2009 version, which they describe as "slightly simplified", had heart rate bands of 55–75, 75–85, 85–90, 90–95 and 95–100% of maximum. The current Olympiatoppen scale (version 2, 2024) uses about 55–72, 72–82, 82–87, 87–92 and 92–100% for its five aerobic zones, and has three further anaerobic zones with no heart rate values. So even "the Norwegian scale" depends on the edition.

• What the extra zones are for. Splitting the range below LT1 separates a gentle jog from a steady aerobic run, and splitting the range above LT2 separates longer hard efforts (about 10 km pace) from shorter ones (3–5 km pace).
• How it maps to three zones. Zone 3 of 5 "coincides well" with Zone 2 of 3: Z1–2 of 5 are below LT1 and Z4–5 of 5 are above LT2. That is an approximate, group-level correspondence.
• Reproducible is not the same as physiological. Anyone can set a repeatable cut at 72% of maximum. That does not mean anything changes in the body there. The scale's authors say the five aerobic zones "are not based on clearly defined physiological markers", and that the additional demarcations are "arbitrary and not definitionally anchored in specific physiological or psychometric transitions". Olympiatoppen describes the transitions between zones as fluid.
• The percentages are from elite athletes and are likely to overestimate the threshold range for beginner and recreational athletes.

A Fellrnr recommendation

If you want a single five-zone scale, I recommend building it on your own thresholds: keep LT1 and LT2 as two of the four cuts. One option, used in the figure, is:

Zone	Name	Range	Worked runner (bpm)	Three-zone equivalent
Z1 of 5	Recovery	Lower part of the range below LT1	Below 130	Z1 of 3
Z2 of 5	Easy	Upper part of the range below LT1	130–145	Z1 of 3
Z3 of 5	Aerobic	Between LT1 and LT2	145–165	Z2 of 3
Z4 of 5	Threshold	Just above LT2	165–173	Z3 of 3
Z5 of 5	Maximum	Near maximum	Above 173	Z3 of 3

• How the other two cuts were chosen. 130 bpm is about 72% of this runner's maximum, the boundary between the first two zones of the Olympiatoppen scale, and about 90% of their LT1 heart rate. 173 bpm is roughly halfway between LT2 and maximum. Both are illustrative coaching choices; I found no physiological reason for either split, and the literature calls such splits arbitrary.
• What this buys you. Any Z1–2 of 5 run is below LT1 and any Z4–5 run is above LT2, so your five-zone log translates directly into the three-zone model used in research, and "Zone 2 of 5" ("Easy") always means running just below LT1. An expert panel recommended that "Zone 2 training" be done immediately below the first threshold.
• Why these names. They follow Garmin's names for Zones 2–5, but Zone 1 is "Recovery" rather than "Warm Up", because very easy running is useful in its own right (see What zones are called).
• No lab test? Estimate LT1, and roughly LT2, with the talk test instead (see Using the talk test to set your heart rate zones).
• What it does not claim. The number of zones and where the subdivisions go are editorial decisions. Two cuts between the thresholds and one above LT2 would be equally defensible, and may suit marathon training better. No trial has compared training outcomes between zone layouts, so there is no evidence that this one produces better results. Fellrnr's editorial preference.
• Generic percentages can clash with your thresholds. The Olympiatoppen 92% boundary would be 166 bpm for this runner, almost exactly their LT2, not a point above it. Measured thresholds take priority.

Watch defaults: Garmin and Polar

Garmin. Garmin's manuals publish one standard table: zones 1–5 at 50–60, 60–70, 70–80, 80–90 and 90–100% of maximum heart rate. The table is the same in manuals from the original Fenix to current models. By default the watch sets the zones from your profile as a percentage of maximum heart rate, with a default maximum of 220 minus age on older models. What differs between models is the choice of basis: all let you edit zones in bpm or percentage of maximum, many add percentage of heart rate reserve, and higher models add percentage of lactate threshold heart rate and separate zones for each sport profile.

Polar. Polar divides the range from 50 to 100% of maximum heart rate into five zones of ten percentage points each: very light, light, moderate, hard and maximum. The manuals checked (Loop, M460, V800, Grit X) use the same table, with maximum heart rate defaulting to 220 minus age, and say the limits can be edited for each sport profile in the Polar Flow web service.

The same colours can change meaning. Change the basis from percentage of maximum to percentage of reserve, switch sport profile, or let the watch detect a new maximum, and the bpm behind each colour moves. A plan written for one setting gives a different instruction on another. Manufacturer documents tell you what the settings are; they are not evidence that the bands match your physiology.

How five-zone definitions differ

The table shows how much "five zones" varies. The last column is the zone for the worked runner at 140 bpm.

Source	Anchor	Z1	Z2	Z3	Z4	Z5	140 bpm is
Garmin and Polar defaults	% maximum	50–60	60–70	70–80	80–90	90–100	Z3 of 5
Seiler and Tønnessen 2009	% maximum	55–75	75–85	85–90	90–95	95–100	Z2 of 5
Olympiatoppen 2024, aerobic zones	% maximum (guidance)	55–72	72–82	82–87	87–92	92–100	Z2 of 5
Runners in the practitioner survey, mean lower limits	% maximum	from 55	from 70	from 80	from 87	from 92	Z2 of 5
MyProCoach, as listed by TrainingPeaks	% maximum	68–73	73–80	80–87	87–93	93–100	Z2 of 5
Fellrnr custom example	Measured thresholds, bpm	under 130	130–145	145–165	165–173	over 173	Z2 of 5

Other systems that a runner may meet:

• Strava has five heart rate zones (Endurance, Moderate, Tempo, Threshold, Anaerobic) calculated from maximum heart rate, which defaults to 220 minus age; its help page does not publish the percentages.
• Apple Watch shows five zones calculated automatically from your health data and lets you edit them; Apple's guide does not publish the boundaries.
• COROS offers three models (lactate threshold, heart rate reserve, maximum heart rate), each with six zones. In its threshold model, Zone 2 is 80–90% of threshold heart rate.
• Suunto: I could not confirm the current default percentages from a first-party source. Older Suunto watches used three zones (60–70, 70–80 and 80–90% of maximum).
• TrainingPeaks lists dozens of scales with three to ten zones.
• Self-selected boundaries vary most at the bottom. In the survey, runners' Zone 2 started anywhere around 70 ± 7% of maximum, and cyclists set their lower three zones lower than runners.

What zones are called

Systems differ in what they call their zones as well as where they put them, and a name suggests what a zone is for. In Romeo and Juliet (Act 2, Scene 2), Juliet asks: "What's in a name? That which we call a rose / By any other name would smell as sweet." The effort of a run does not change with its label, but what a runner does with that run often does. (80/20 Endurance borrowed the line for an article on its own zones.)

Source	Basis	Zone 1	Zone 2	Zone 3	Zone 4	Zone 5	Beyond zone 5
Garmin	% of maximum heart rate	Warm Up	Easy	Aerobic	Threshold	Maximum	–
Polar	% of maximum heart rate	Very light	Light	Moderate	Hard	Maximum	–
Strava	Maximum heart rate	Endurance	Moderate	Tempo	Threshold	Anaerobic	–
COROS, threshold model	% of threshold heart rate	Recovery	Aerobic Endurance	Aerobic Power	Threshold	Anaerobic Endurance	6: Anaerobic Power
Olympiatoppen 2024, effort descriptions	% of maximum heart rate (guidance)	Very light	Fairly light	Somewhat hard	Hard	Very hard	6–7: Very, very hard; 8: Maximal mobilization
Research three-zone model	Measured thresholds	Low	Moderate (the threshold region)	High	–	–	–
Metabolic domains	Measured thresholds and critical power	Moderate	Heavy	Severe	–	–	–
Coggan, cycling power	% of FTP	Active Recovery	Endurance	Tempo	Lactate Threshold	VO2 Max	6: Anaerobic Capacity; 7: Neuromuscular Power
Stryd, running power	% of threshold power	Easy	Moderate	Threshold	Interval	Repetition	–
Daniels, running paces (see Jack Daniels)	Paces from a race result	Easy	Marathon	Threshold	Interval	Repetition	–
Fellrnr five-zone scale (this article and Training Budget; see my recommendation)	Your own LT1 and LT2 plus two chosen cuts	Recovery	Easy	Aerobic	Threshold	Maximum	–

Apple numbers its five zones without naming them, Friel's TrainingPeaks guide numbers its seven (1, 2, 3, 4, 5a, 5b, 5c), and the 80/20 Endurance system uses 1, 2, X, 3, Y, 4, 5, where X and Y are ranges to avoid.

The Fellrnr names. The Fellrnr five-zone scale uses Garmin's names for Zones 2 to 5, so the names on a Garmin watch line up with this article's zones once the watch is set to your own boundaries. Zone 1 is called "Recovery" rather than "Warm Up" because running below Zone 2 is useful in its own right, for recovery runs and very easy volume, not just for warming up. Because each zone is anchored to your thresholds, the names describe the purpose of the zone: "Easy" is running below LT1, "Aerobic" is sustained work between LT1 and LT2, and "Threshold" starts at your measured LT2. The names are an editorial choice, not a physiological finding.

How names steer what you do:

• The same word, opposite meanings. "Moderate" is the easiest range in the metabolic-domain scheme, where the moderate domain lies below the first threshold. It is the middle range, between the thresholds, in the research three-zone model. It is Polar's Zone 3 (70–80% of maximum) and Strava's Zone 2. A runner told to "keep it moderate" could be jogging easily or running just below LT2.
• "Threshold" that is below threshold. Garmin calls Zone 4 "Threshold", but its default Zone 4 is 80–90% of maximum heart rate: 144–162 bpm for the worked runner, below their LT2 of 165. A "threshold" session run to the watch's label would sit between this runner's thresholds, not at LT2. COROS ("Threshold", 95–102% of threshold heart rate) and Coggan ("Lactate Threshold", 91–105% of FTP) anchor the name to a threshold, so the name and the effort match more closely. The Fellrnr scale uses Garmin's word but starts the zone at the runner's measured LT2, which is the point of this section: the same name can mean different efforts depending on where the boundaries are.
• "Endurance" moves. It is Strava's Zone 1, part of COROS's Zone 2 ("Aerobic Endurance") and Coggan's level 2.
• "Aerobic" describes a purpose, not the only aerobic zone; "anaerobic" misleads. Almost all running is mainly aerobic: in all-out efforts, aerobic and anaerobic energy contribute equally at around 75 seconds, and the aerobic share is larger in anything longer. So when Garmin, and the Fellrnr scale, call Zone 3 "Aerobic", it means the zone for developing aerobic fitness with sustained work, not that the other zones are less aerobic. Strava's "Anaerobic" Zone 5 is more misleading: "anaerobic" describes efforts of seconds, not a 5-minute repetition that reaches Zone 5 heart rate.
• Names come with a promised benefit. Garmin's manual lists a "Benefits" column, such as "Improved aerobic capacity, optimal cardiovascular training" for Zone 3 and "Beginning-level aerobic training, reduces stress" for Zone 1, and Polar lists "Benefits" and "Recommended for" by zone. These are manufacturers' descriptions, not trial results. With Garmin's defaults, the worked runner's genuinely easy run at 140 bpm is labelled "Aerobic" and "optimal cardiovascular training", and anything below 108 bpm is labelled "Warm Up". Neither label tells the runner that 140 is below their first threshold. On the Fellrnr scale, built on this runner's thresholds, the same run is Zone 2 of 5, "Easy".
• "Zone 2" is now a name in its own right. "Zone 2 training" usually means running just below the first threshold, the meaning an expert panel recommended. On default watch settings, Zone 2 is 60–70% of maximum heart rate, well below LT1 for most runners.

Practical rule. Read the boundaries, not the name. Fellrnr.com names a zone by its number and system ("Z2 of 5") and, where it matters, by its boundaries ("below LT1"). For its own five-zone scale it also uses the names Recovery, Easy, Aerobic, Threshold and Maximum; other descriptive names are the source's own.

Percentage of what? Maximum, reserve or threshold

Most of this article uses percentages of maximum heart rate, because that is what watches use by default. Two other bases are common, and each moves every boundary. For the worked runner, 70% is 126 bpm of maximum, 141 bpm of reserve and 116 bpm of a threshold heart rate of 165.

Basis	You need	Target heart rate	Strengths	Weaknesses
% of maximum heart rate	Maximum	fraction × maximum	One number; the default on most watches; simple arithmetic	Ignores resting heart rate; the thresholds sit at different percentages in different runners; an age-estimated maximum adds a typical error of about 11–12 bpm
% of heart rate reserve (the Karvonen method)	Maximum and resting heart rate	resting + fraction × (maximum − resting)	Matches percentage of oxygen uptake reserve almost one for one; allows for a low resting heart rate	Two numbers to get right; resting heart rate depends on how and when it is measured; the thresholds still sit at different percentages
% of threshold heart rate (LTHR)	Heart rate at the second threshold, from a laboratory test or a time trial	fraction × LTHR	Anchored to a threshold rather than the maximum; no maximal test needed; zones follow fitness if you retest	Needs a hard test, repeated; the test differs between systems; the position of LT1 is still assumed
Measured thresholds	Heart rate at LT1 and LT2	Set the boundaries in bpm directly	Both physiological boundaries are your own	Needs laboratory or careful field testing; results depend on the method

• Reserve tracks oxygen uptake better than maximum does. In 63 adults on a cycle ergometer, percentage of heart rate reserve matched percentage of oxygen uptake reserve almost exactly (slope 1.00, intercept −0.1), but not percentage of V̇O₂max (slope 1.12, intercept −11.6); the authors concluded that "%HRR should not be considered equivalent to %VO2max". Percentage of maximum heart rate is offset further: in 162 young adults, 40% of V̇O₂max came at 63% of maximum heart rate (see How effort rises with heart rate).
• But no percentage locates the thresholds. In 863 people, from sedentary adults to trained athletes, the first ventilatory threshold was at 55.5 ± 12.1% of heart rate reserve on average, with group averages from 49.5% in sedentary adults to 67.8% in aerobic athletes; in oxygen uptake reserve, individual values ranged from 25.5% to 86.0%. In 165 recreational runners, no fixed percentage of any maximum placed everyone in the same intensity domain.
• Resting heart rate is a second source of error. Its value depends on when and how it is measured (on waking, sitting, or a watch's overnight average), so measure it the same way each time. In a reserve-based target, an error in resting heart rate moves a 70% boundary by 0.3 times the error, and an error in maximum moves it by 0.7 times the error.
• Threshold-based prescription did better in training studies. An individual-participant meta-analysis of 1,544 people in 42 studies found larger gains in V̇O₂max when intensity was set from measured thresholds than from percentages of maximum heart rate, heart rate reserve or V̇O₂max (4.1 versus 1.8 mL/kg/min in controlled studies), but no evidence that the spread of individual responses was smaller. Those studies used laboratory thresholds, not a percentage of a field-test threshold heart rate.
• Threshold heart rate still assumes where LT1 is. Friel's TrainingPeaks running zones put the top of Zone 2 at 89% of LTHR. That fits group averages: LT1 heart rate is about 89–91% of LT2 heart rate in recreational runners (about 80% and 90% of maximum) and in habitual runners (85.1% and 93.5% of peak). Individuals vary around those averages, so for LT1 a percentage of threshold heart rate is still an estimate. Its advantage is the starting point: fixed anchors estimate LT2 more accurately than LT1. Inference.
• FTP and critical power anchor power, not heart rate. Functional threshold power and critical power are power thresholds, used for power zones (see Power zones); heart rate systems use the heart rate measured at a threshold instead. Friel's TrainingPeaks test takes the average heart rate over the last 20 minutes of a 30-minute solo time trial; his revised guide uses a 20-minute time trial minus 5%. These give different numbers, so check which one a plan uses.

Which to use. If you have measured thresholds, set your zones in bpm around them. If you have a reliable threshold heart rate from a time trial, percentages of it are the next best choice. Between maximum and reserve, reserve tracks oxygen cost better, but neither tells you where your thresholds are. Fellrnr's view, based on the evidence above.

Friel and other threshold-based coaching systems

Joe Friel uses seven heart rate bands (1, 2, 3, 4, 5a, 5b, 5c) based on threshold heart rate. Which version matters. In the version on TrainingPeaks, the running zones are below 85%, 85–89%, 90–94%, 95–99%, 100–102%, 103–106% and above 106% of LTHR, with different percentages for cycling, and LTHR comes from a 30-minute time trial. In his own guide, revised in September 2025, he writes "I no longer separate running and cycling heart rate zones by differing ranges", estimates threshold heart rate from a 20-minute time trial minus 5%, and places LT1 at the junction of zones 2 and 3 and LT2 at the border of zones 3 and 4. The revised percentages are published as an image that I could not read, so they are not quoted here. For the worked runner, with an LTHR of 165, 140 bpm is 84.8% of LTHR: the very top of Zone 1 of 7 in the TrainingPeaks running version. Matt Fitzgerald's 80/20 system also uses seven zones (1, 2, X, 3, Y, 4, 5) based on threshold heart rate, with Zone 1 starting at 72% of LTHR. Do not mix definitions from different versions or systems.

Power zones are different

Coggan's classic power levels for cycling are seven levels defined as percentages of functional threshold power (FTP), the power a trained rider can hold for about an hour, often estimated as 95% of the average power in a 20-minute all-out effort:

Level	Name	% of FTP
1	Active Recovery	Below 55
2	Endurance	56–75
3	Tempo	76–90
4	Lactate Threshold	91–105
5	VO2 Max	106–120
6	Anaerobic Capacity	121 and above
7	Neuromuscular Power	Maximal

• Three different reference points. FTP is a field estimate of the power you can sustain for about an hour. Critical power is the boundary between the heavy and severe domains, estimated from several maximal efforts. Power at maximal oxygen uptake is higher than both. They are not interchangeable.
• Heart rate is a response; power is an output. Heart rate has a ceiling and a delay. Cycling watts measure the work being done at that instant. A five-second sprint can produce far more power than a several-minute effort that brings oxygen uptake to its maximum, while heart rate is still climbing.
• There is no "steady-state power at maximum heart rate". At a high enough constant power, no metabolic steady state is possible: oxygen uptake and heart rate keep rising until you stop. Many different powers end at the same maximum heart rate. The TrainingPeaks version of Coggan's table notes for level 6 that "Heart rate generally not useful as guide to intensity", and the top power levels should not be translated into heart rate percentages.
• Running "power" is an estimate. Watches and footpods calculate running watts from speed, gradient and movement. They are not equivalent to watts measured at a bicycle crank (see Running Power Meters).

Where does Zone 1 begin?

This is three separate questions.

What counts as exercise? Exercise is usually defined by its purpose: physical activity that is "planned, structured, and repetitive" and aimed at improving or maintaining fitness. By that definition a slow walk done for training is exercise at any heart rate. In the three-zone model, Z1 of 3 is simply "exercise below LT1" and has no lower limit.

What will my watch count as Zone 1? Whatever its settings say. Polar and Garmin defaults start Zone 1 at 50% of maximum heart rate, which is 90 bpm for the worked runner; anything lower is not in a zone. Olympiatoppen starts its first zone at about 55%. In the practitioner survey, runners set the bottom of Zone 1 at 55 ± 10% of maximum on average and cyclists at 47 ± 18%. These are logging conventions that describe practice. None is a biological minimum.

How easy can exercise be and still improve something? It depends on the outcome, the person and the amount.

• Cardiorespiratory fitness in the general population. A review of training studies concluded that the evidence for a threshold intensity "was not strong", and suggested about 30% of oxygen uptake reserve as a minimum for people with a starting VO₂max below 40 mL/kg/min and about 45% of oxygen uptake reserve for those above it. Those figures are in oxygen uptake reserve, not percentage of maximum heart rate, and the "higher fit" group was far below the fitness of competitive runners. Limited evidence.
• Fitter people need more. The same review's pattern, a higher minimum for fitter subjects, suggests that very easy exercise that improves an unfit person may only maintain a trained runner. How easy a trained runner can go and still gain is not established.
• Duration and total dose matter. Elite endurance athletes do most of their low-intensity training at about 60–65% of maximal oxygen uptake, in sessions lasting hours. The survey's authors suggest that cyclists set lower zone boundaries because their easy sessions are much longer. Consistent elite practice, not a trial result.
• Other outcomes. Recovery, time on feet and health may respond to lighter activity than aerobic fitness does (see Recovery Runs). This article does not review the health evidence.

A low reading on the watch therefore says little about whether a long walk or a very easy run was worthwhile. If you need a rule for your log, pick a floor (for example, count everything done as training) and apply it consistently.

How effort rises with heart rate

Heart rate rises roughly in a straight line with the oxygen you use, but blood lactate, how long you can keep going and how long you take to recover all change much more steeply near the top. That is why training-load methods weight a minute at a high heart rate far more heavily than a minute at a low one. This section is about that relationship, not about zones.

The cost of a minute, relative to a minute at 70% of maximum heart rate. Oxygen uptake is from Swain and colleagues' regression; the Banister lines use a maximum of 180 and a resting heart rate of 50 bpm; the doubling rule is the Training Budget's. At 90% of maximum, a minute uses about 1.6 times the oxygen of a minute at 70%, but counts 2.3–2.5 times as much on the lactate-based weightings.

Different measures of effort

Measure of effort	How it changes as heart rate rises	Evidence
Oxygen uptake and energy use	Close to a straight line	In 162 young adults, 40, 60, 80 and 85% of V̇O2max came at 63, 76, 89 and 92% of maximum heart rate
Perceived effort	Roughly a straight line in steady exercise	The Borg 6–20 scale was built to rise in step with heart rate; ratings averaged 10.8 at LT1 and 13.6 at LT2
Blood lactate	Flat, then rising, then rising steeply	About 1.0–2.0 mmol/L at LT1 and 2.5–4.0 mmol/L at LT2 in trained athletes, with a second, steeper rise at LT2
How long you can keep going	Falls steeply	For well-trained runners, Olympiatoppen gives about 3 hours at 72–82% of maximum heart rate, 60 minutes at 82–87%, 30 minutes at 87–92% and 15 minutes above 92%
Recovery afterwards	A step at the thresholds	After 60 or 120 minutes below VT1, heart rate variability returned to its pre-training level within 5–10 minutes; threshold and harder sessions delayed recovery much more

Established formulas

• Oxygen uptake. Swain and colleagues' regression is %HRmax = 0.64 × %V̇O₂max + 37, so %V̇O₂max ≈ (%HRmax − 37) ÷ 0.64. At 70% of maximum heart rate that is about 52% of V̇O₂max; at 90%, about 83%. With heart rate reserve the match is closer still: percentage of reserve ≈ percentage of oxygen uptake reserve (see Percentage of what?).
• Banister's TRIMP weights each minute by x × A × e^Bx, where x is the fraction of heart rate reserve, with A = 0.64 and B = 1.92 for men and A = 0.86 and B = 1.67 for women. The exponential term follows the typical rise of blood lactate with heart rate, using group data (see TRIMP).
• Individualised TRIMP replaces Banister's group constants with an exponential fitted to the runner's own heart rate and lactate test. In eight recreational distance runners over 8 weeks, weekly individualised TRIMP was related to gains in speed at 2 and 4 mmol/L (r = 0.87 and 0.74) and to 5,000 m and 10,000 m performance (r = −0.77 and −0.82), whereas group-average values showed no significant relationship. Limited evidence: one small study.
• Edwards multiplies time in 10-point bands from 50–60% to 90–100% of maximum heart rate by 1 to 5: a straight-line weighting in steps that stops rising at 90%.
• Lucia multiplies time below VT1, between VT1 and VT2, and above VT2 by 1, 2 and 3, as the method is usually described.
• Fellrnr's doubling rule weights a minute by 2^{(%HRmax − 78) ÷ 15}: the weight doubles for every 15 percentage points of maximum heart rate. It is a rounded form of Banister's men's curve, scaled so that easy running at about 78% of maximum, in the Easy zone, counts as 1 (see Training Budget). In the chart, like the other methods, it is shown relative to a minute at 70%.

What this means. Heart rate is a good guide to oxygen cost but understates how much harder the top end is. Going from 80% to 90% of maximum adds about 24% to the oxygen cost of a minute, but 50–58% to its weight on the lactate-based scales. None of these curves is individual: Banister's constants come from group data, and individual lactate curves differ. Heart rate itself also drifts upwards during long or hot runs at the same effort (see Heart Rate Drift). Strong evidence for the near-linear oxygen relationship; moderate for the exponential load weightings, which are models built on group lactate data.

Why the curves have no bend at the thresholds

If LT1 and LT2 are turning points, you might expect the curves above to bend at them. They do not, mainly because they are drawn against percentage of maximum heart rate, and different runners reach their thresholds at different percentages. Four lines of evidence bear on this:

• Each runner's bend is in a different place. In 1,411 endurance-trained runners, heart rate at the first ventilatory threshold averaged 85.1 ± 4.6% of peak heart rate, and at the second 93.5 ± 2.5%. In 165 recreational runners, "there was no fixed intensity that would have allowed all individuals to be exercising within the same domain". A formula fitted to group data averages bends that sit at different heart rates, and the average of many bends in different places is a smooth curve. This is my explanation from the spread of thresholds; the formulas' authors did not test it.
• Lactate may rise smoothly rather than turning sharp corners. In 23 ramp tests, a smooth exponential described blood lactate better than a threshold model in 21, and the authors concluded that the rise "could be better described mathematically by a continuous model". On that view, LT1 and LT2 are useful markers on a curve that steepens smoothly, not kinks. This is one study, and threshold-based methods remain standard in research and coaching (see three-zone model).
• Heart rate often bends too, near LT2 (see Heart Rate Deflection). In 227 young men in a cycle ramp test, 85.9% showed a "regular" downward bend in their heart rate curve, 6.2% showed no bend and 7.9% an inverted (upward) bend. Where a bend appeared, the power at it did not differ significantly from the power at the second lactate turn point. A review found that "a high degree of relationship exists between HRDP and the second lactate turnpoint", but also that "the degree of HR deflection is highly dependent upon the type of protocol used". A downward bend means each beat above LT2 stands for more work than a beat below it, so a curve drawn against heart rate would, if anything, understate the top end further (my inference).
• The clearest break shows up over time, not in a ramp. Below critical power, oxygen uptake and lactate settle at a constant pace; above it they keep rising (see below). A formula that weights each minute by its heart rate alone cannot see how long you have been above that boundary.

Formulas that do have a bend.

• Lucia's TRIMP steps at VT1 and VT2 by construction, weighting time by 1, 2 and 3. It needs measured thresholds, and every minute within a zone counts the same.
• Individualised TRIMP fits an exponential to the runner's own heart rate and lactate test, so the steep part falls where that runner's lactate rises. It is still smooth, without a corner, but the steepening is in the right place for that runner.
• W′ balance, used with power, changes direction at critical power: "Power output above CP results in depletion of the W'", and the reserve is rebuilt below critical power, faster the further below it the runner or rider is. It is the only model here that tracks time spent above the boundary, but it uses power rather than heart rate and was developed in seven cyclists.

I found no heart rate formula that bends at both thresholds and has been tested in runners. Moderate confidence: my search was thorough but not a systematic review.

What this means. Treat the smooth curves as averages. For an individual runner, the cost of a minute probably rises more steeply around their own thresholds than the curves show, especially at LT2, and above LT2 it also depends on how long they stay there. Recovery shows a similar step: heart rate variability recovered within minutes after long sessions below VT1, but much more slowly after threshold and harder sessions. Inference, moderate confidence.

Above LT2, the same pace gets harder

Below LT2, the oxygen cost of a steady pace or power levels off, given time. Above it, oxygen use and heart rate keep climbing until you slow down or stop. Strictly, the boundary is critical power (or critical speed) and the maximal lactate steady state, which lie close to LT2 but are not identical to it (see Lactate Threshold).

• Below LT1 (the moderate domain): "During moderate‐intensity exercise, V̇O2 and blood [lactate] reach steady‐state values rapidly".
• Between LT1 and LT2 (the heavy domain): an extra, slowly developing oxygen cost appears, called the slow component, and "the attainment of (higher amplitude) steady‐state values for V̇O2 and blood [lactate] is delayed".
• Above critical power (the severe domain): "V̇O2 and blood [lactate] continue to rise until" maximal oxygen uptake is reached. The review does not report heart rate, but heart rate rises with oxygen uptake, so at a constant pace above this boundary heart rate climbs towards maximum. Strong evidence for oxygen and lactate; the heart rate statement is an inference.

Is lactate build-up the cause? Probably not. Lactate rises alongside the drift, but it is better read as a sign that you are above the boundary than as the cause of the extra effort:

• The slow component "represents a progressive loss of skeletal muscle contractile efficiency and is associated with the fatigue process", and is linked to the recruitment of additional, less efficient type II muscle fibres. The muscles need more oxygen for the same pace, so heart rate rises.
• Jones and colleagues write that blood lactate, per se, "is neither an appropriate nor a sufficiently sensitive metric to enable a confident assessment" of the steady state, and that a steady lactate "need not imply the existence of a bioenergetic steady‐state in contracting skeletal muscle".
• A review of lactic acid and fatigue concluded: "It should not be taken as fact that lactic acid is the deviant that impairs exercise performance".

Moderate confidence on the mechanism: these are reviews, and the causes of fatigue are still debated.

Heart rate also drifts below LT2, for a different reason. In long or hot sessions, heart rate rises at the same pace even when the effort is easy. In one cycling study at 60% of V̇O₂peak, heart rate rose by 18.5 bpm between 20 and 50 minutes, and Coyle and González-Alonso describe this cardiovascular drift as occurring as body temperature rises (see Heart Rate Drift). Below LT2 the oxygen cost of the pace eventually settles while heart rate drifts; above LT2 the oxygen cost itself keeps rising.

In practice. A heart rate cap works for easy runs, and for steady work just below LT2 once heart rate has settled. Above LT2, heart rate is a poor target: it lags at the start of each effort and then keeps climbing while the pace stays the same, so holding a fixed heart rate would mean slowing down. Use pace, power or perceived effort for these sessions, and expect heart rate to rise (see When heart rate helps).

The worked example as a table

The same information as the figure, for a runner with a maximum of 180, LT1 at 145 and LT2 at 165 bpm. Cells are coloured by zone number, using paler versions of Garmin's zone colours: 1 grey, 2 blue, 3 green, 4 orange, 5 red. The colour shows the label, not the effort: Z3 of 3 is the hardest zone of its system but shares green with Z3 of 5.

Heart rate (bpm)	Three zones (measured thresholds)	Custom five zones	Polar default five zones
Below 90	Z1 of 3 (if exercising)	Z1 of 5	No zone
90–108	Z1 of 3	Z1 of 5	Z1 of 5
108–126	Z1 of 3	Z1 of 5	Z2 of 5
126–130	Z1 of 3	Z1 of 5	Z3 of 5
130–144 (includes 140)	Z1 of 3	Z2 of 5	Z3 of 5
144–145	Z1 of 3	Z2 of 5	Z4 of 5
145–162	Z2 of 3	Z3 of 5	Z4 of 5
162–165	Z2 of 3	Z3 of 5	Z5 of 5
165–173	Z3 of 3	Z4 of 5	Z5 of 5
173–180	Z3 of 3	Z5 of 5	Z5 of 5

Boundary types: 145 and 165 are measured for this runner; 130 and 173 are illustrative; 90, 108, 126, 144 and 162 are Polar's published 50, 60, 70, 80 and 90% of a maximum of 180.

Confidence summary

Claim	Type	Confidence
Only the two thresholds have a physiological basis; other zone boundaries are conventions	Finding (stated by the scale's own authors)	High
No fixed percentage of maximum heart rate places all runners in the same zone	Finding	High
Five-zone and three-zone scales are the most used	Finding (one self-selected survey)	Moderate
Zone 3 of 5 on the Norwegian scale corresponds to Zone 2 of 3	Finding (approximate, group level)	Moderate
Garmin and Polar default to five ten-point bands from 50% of maximum	Manufacturer documentation	High
Age equations are right on average and often wrong for individuals	Finding (cycle test, sedentary adults)	High for the pattern; moderate for runners
The talk test marks roughly the first threshold without any measured numbers	Finding (several small studies, mostly young healthy adults)	Moderate
The talk test's "no" stage marks roughly the second threshold	Finding (one study, 18 trained cyclists)	Low to moderate for runners
For runners without test results, the talk test is a better guide to easy running than an age-estimated heart rate zone	Inference from the findings above	Moderate
Heart rate zones set from a talk test are better than zones calculated from a formula	My opinion; not tested directly	Untested
Chest straps agree with an ECG more closely than wrist-worn optical sensors, and optical sensors can lock on to cadence	Finding (50 adults; studies of optical sensor error)	High
An optical sensor on the upper arm closely matched a chest strap across activities	Finding (one validation study)	Moderate
ECG straps are much better than optical sensors, the Polar Verity Sense is good enough for most work, and a low strap battery increases errors	My experience, consistent with the studies and with Polar's guidance	Moderate
Automatic maximum heart rate detection tends to pick up bad data	My experience; Garmin confirms the feature is on by default	Untested
Percentage of heart rate reserve matches percentage of oxygen uptake reserve	Finding (63 adults, cycle)	High
No basis (maximum, reserve or threshold heart rate) places every runner's LT1 correctly	Finding	High
Prescribing from measured thresholds gives larger V̇O2max gains than fixed percentages	Finding (individual-participant meta-analysis)	Moderate
Zone names differ between systems and suggest different purposes	Manufacturer and coaching documentation	High
The names Recovery, Easy, Aerobic, Threshold and Maximum for the Fellrnr five-zone scale	Editorial choice	Not a factual claim
Oxygen uptake rises roughly in a straight line with heart rate; lactate-based load rises exponentially	Finding and model	High for oxygen; moderate for the load weightings
Group formulas show no bend at LT1 or LT2 mainly because runners reach their thresholds at different percentages of maximum	Inference from the spread of thresholds	Moderate
Blood lactate in a ramp test is described better by a smooth curve than by a threshold	Finding (one study of 23 tests)	Moderate; the threshold view is still widely used
Most people show a bend in heart rate near LT2 in a ramp test, but some show none or an upward bend	Finding (227 young men; review)	Moderate; depends on the test protocol
At a constant pace or power above critical power, oxygen uptake and lactate keep rising until maximum; below it they level off	Finding (review)	High
At a constant pace above that boundary, heart rate keeps rising towards maximum	Inference from the oxygen findings	Moderate to high
The rise above LT2 is caused by lactate build-up	Not supported: lactate rises alongside it; falling muscle efficiency and fatigue are the leading explanation	Moderate on the mechanism
A minimum intensity for improving fitness of about 30–45% of oxygen uptake reserve	Finding (review; not in trained runners)	Low for runners
A custom scale that keeps LT1 and LT2 is the most useful five-zone scale	Editorial preference	Untested

See also

• Heart Rate, Maximum Heart Rate, Heart Rate Reserve and Resting Heart Rate
• Heart Rate Drift and Heart Rate Deflection
• Optical Heart Rate Monitoring, Fixing Heart Rate Monitor Problems and Data Quality
• Lactate Threshold, Critical Power and Tempo and Threshold Training
• Rating of Perceived Exertion
• Training Budget, How Often To Run and Recovery Runs
• Pace Range Ratio and Walking Breaks
• TRIMP
• Running Power Meters