The Science of the Long Run

While the Long Run is a core part of training for endurance races, there is relatively little scientific evidence available. This entry summarizes the evidence I have been able to locate.

1 Correlations between The Long Run and Performance

This section summarizes the studies that correlate the long run against various factors including marathon finish time and hitting the wall, or injury rate.

1.1 The Long Run and Marathon Performance

A 1970's study looked for correlations between marathon finish times and metrics from the final two months of training before the race. There is an obvious correlation between the finishing time and both the longest run and overall mileage.

Time	Max Week	Longest run	Number of 20+ mile runs	Mileage over two months
2:25:00	106	22.3	2.7	627
2:38:00	86	20.9	2.7	477
2:54:00	80	21.5	2.9	445
3:08:00	64	20.5	1.9	331
3:23:00	57	18.2	1.1	291
3:38:00	58	18.8	1.2	312
3:53:00	46	17	0.7	224
4:15:00	42	15.3	0.4	208
4:52:00	37	13.1	0.3	174

The length of the longest run against finish time.

The number of 20+ mile long runs against finish time.

Another similar study showed that the number of runs over 16, over 20 miles, the length of the longest run and overall mileage are correlated with finish time. One study separated out overall mileage from the long run. In that study, two groups of runners did the same long run schedule but their overall weekly mileage either increased from 18 miles to 39 miles, or from 23 miles to 49 miles. The marathon performance was identical between the groups, suggesting that long run distance may be more important than overall mileage.

1.2 The Long Run and Hitting the Wall

A study of 315 marathon runners evaluated the factors that are correlated with reported 'Hitting the Wall'. The length of the longest run but not weekly mileage was correlated hitting the wall. Further analysis showed that a longest long run of 20 miles or less increased the chance of hitting the wall by 50%.

1.3 Injuries

I found no studies that showed a correlation between the length of the long run and injuries. There is some evidence that the increase in mileage, overall monthly mileage or both is correlated with injuries.

1.4 Correlation is not Causation

What little scientific evidence we have correlates better marathon outcomes with longer distance long runs. However, the evidence does not indicate if the longer distance long runs cause the improved marathon outcome or if better runners tend to do longer runs.

XKCD's explanation of correlation and causation, from [1], also available as a Tee Shirt (I want one).

2 Glycogen Depletion

The evidence from the Glycogen depletion patterns may give some insight into the long run. We see that glycogen is depleted from some fibers before others. This pattern suggests that some slow twitch fibers are used first, and as these become exhausted, other fibers are recruited in their place. Slow twitch fibers are used first, but over time more and more fast twitch fibers are recruited. This may be one mechanism behind the benefits of the long run; it exhausts the most accessible fibers and therefore trains the other fibers. It is possible that without the long run, only the same, easily recruited fibers are trained. It is also conceivable that by using fast twitch fibers for endurance exercise, the long run will change the nature of these fibers.

3 Muscle Damage

Running, and especially Downhill Running tends to produce muscle damage and Delayed Onset Muscle Soreness. This damage immediately weakens the muscles, with recovery and remodeling of the muscle fibers taking around 14 days. After a single bout of DOMS, the muscle fibers undergo 'profound adaptations' to be more resilient. It seems reasonable that this mechanism is a key benefit of the Long Run. Therefore a long run should be long enough to create some muscle damage, while balancing the risk of injury.

• Muscle damage from eccentric exercise (downhill running)
• 

  Muscle before downhill running

• 

  Immediately after downhill running. Notice the disruption to the dark bands (z-bands) that are part of the muscle structure showing there is immediate damage.

• 

  One day later, the damage and disruption is worse, indicated some continued breakdown.

• 

  Muscle 14 days later, structurally recovered (other metrics do not return to pre-exercise levels at 14 days).

4 Anecdotal Advice

Given the limited scientific information, it seems reasonable to examine the anecdotal advice that is available.

• There are instances of remarkable performances without running long in training. For instance, Grete Waitz never ran more than 12 miles in training before winning the 1979 New York Marathon. However, at the time she was a world class track and cross country athlete who trained at 80-90 miles per week. Her book on marathon training emphasizes the importance of the long run.
• Most training plans peak at 20 miles, though JD has 22 mile runs and Jeff Galloway has a 26 mile (2 min/mile slower with walking breaks)
• A consensus of caches suggests a single 20 mile run for novice marathoners and 3-6 long runs peaking at 23 miles for the more experience. .
• Some coaches recommend limiting the length of the long run to a percentage of the weekly mileage, often in the range 25-35%. The rationale for this unclear, and this recommendation could encourage high levels of Training Monotony.
• Using time rather than distance for long runs is sometimes suggested, as a 16 mile run at 11 min/mile would take a similar time to 22 miles at 8 min/mile. For instance, Jack Daniels does not recommend novice marathoners to do a long run longer than 2.5 hours. This approach may be based around the belief that runs longer than 2 or 3 hours provide little or no additional benefit. While there are some animal studies that show the benefits of endurance training plateaus, these studies looked at a limited aspect of endurance (cytochrome c) and other animal studies do not show this plateau.
• It is often noted that ultramarathon runners do not train with proportionately long training runs. So a runner training for a 100 mile race will not run 76 mile long runs, which would be the equivalent to a 20 mile run for a marathon. In fact, the great ultrarunner Ray Krolewicz does not do training runs longer than 22 miles. However, ultrarunners often run many long runs close together. For instance, Ray has done more than one 22 miler on each day of the weekend, and 'back to back' long runs of 20-30 miles are commonly prescribed. Also, ultrarunners typically race frequently enough for the races to contribute to their overall training.
• It seems that many runners who run over 24 miles often do so in organized events, even if they are using the event for training rather racing.

5 Recommendations

Here is my personal interpretation of the scientific and anecdotal evidence.

• For marathon running, the longest long runs should be over 20 miles.
• There may be some injury risk to longer long runs, but it seems more likely that the injury risk comes from ramping up the length of the long run too quickly.
• Looking at the time for Endurance Adaptations to occur, it seems likely that most marathon training plans tend to be too short for many novice runners. Some are as short as four months, but even six months may be too short for many runners.
• The ideal pace for long runs is unclear, but almost all marathon runners train at slower than race pace for their long runs.