How long should your long runs be?
Faster marathoners do longer long runs, and regular runs of 90 minutes or more seem to help you hold your pace late in a race. But this evidence compares different runners, and every popular cap, from 16 miles to 3 hours, is untested convention. Build your long run gradually and let the 10% spike rule set the pace of the build.
What to do
- From the middle of your build, include at least one run of 90 minutes or more each week if you run four or more days a week.
- Aim for a few runs of 32 km (or about 2 hours 45 minutes if you are slower) in the last 10 weeks. Two to five is typical; the number is convention.
- Treat a cap of about 3 hours as a sensible default, not a law. It is coaching convention.
- In the last 8–12 weeks, add marathon-pace sections to some long runs and keep the others easy. This copies elite practice and is untested in recreational runners.
- Whatever the plan says, keep each run within 10% of your longest run of the past 30 days.
What the volume data show
The biggest summary is a meta-regression of 85 studies and 137 groups of runners (Doherty and colleagues, 2020). It linked every marker of training volume to faster marathon times. Those markers included the longest run and the number of runs of 32 km or more, as well as weekly distance, runs per week, peak week and hours per week.
There are two catches. First, the data compare different people. Faster runners may do longer runs because they are fitter, not the other way round. Second, no randomized trial has varied long-run length and measured marathon results. The benefit of volume also seems to level off: in one model of recreational runners, the gain from extra weekly distance shrank as distance rose (Tanda and Knechtle, 2013).
In first-time marathoners, a longer longest run in training was protective against injury, as was having completed a half marathon (Toresdahl and colleagues, 2020). See also our page on run-walk and first marathons.
Durability: why long runs may matter
Durability is how much your fitness markers, such as threshold pace and economy, slip during a long effort. Research now treats it as a separate predictor of marathon performance (Maunder and colleagues, 2021; Jones and colleagues, 2024).
- In 82,303 marathoners, the drift between heart rate and pace first showed up at about 25 km. Runners who drifted less raced closer to their critical speed, a fitness marker estimated from their training (Smyth and colleagues, 2022).
- After a 90-minute run, London marathon finishers’ lactate-threshold speed fell from 12.8 to 12.1 km/h. The bigger the drop, the slower their marathon (Hunter and Muniz-Pumares, 2025).
- Runners who regularly ran 90 minutes or more lost about half as much economy during a 90-minute run as matched runners whose runs were all under 70 minutes: 3.1% against 6.0% (Zanini and colleagues, 2026).
- Women matched for performance lost about 1% of time-trial speed after three hours of running, against about 10% for men (Jaén-Carrillo and colleagues, 2026).
Expert guidance on building durability is to keep volume high over years and to do regular long sessions that include race-pace or progressive sections, plus strength work (Jones and Kirby, 2025). The authors say openly that this rests on scant data, and it is still unclear whether durability can be trained or is mostly a trait (Hunter, 2026). See also strength training for runners.
The popular caps are convention
These long-run rules have no controlled or observational test:
- A cap of 2.5–3 hours per run.
- The Hansons 16-mile ceiling and its idea of “cumulative fatigue”.
- Pfitzinger’s 20–22-mile long runs.
- Fast-finish and marathon-pace long runs.
That does not make them wrong. It means you can pick a style you like. The volume data, if anything, link longer longest runs with faster times, which sits awkwardly with a hard 16-mile cap.
Elite marathoners do a weekly long run of 30–40 km, slightly slower than marathon pace, and treat it as one of their most important sessions (Haugen and colleagues, 2022).
A practical framework
These defaults come from our source report’s rule set. None is better than grade C, and most are convention.
| Rule | What it says | Grade |
|---|---|---|
| VL-4 | From mid-build, at least one run of 90+ minutes a week if you run 4+ days a week. | C |
| VL-5 | Count runs of 32 km or more (or 2:45+ for slower runners). Two to five in the last 10 weeks is typical. | C/D |
| VL-6 | Cap at about 3 hours (2:30–3:00 if you run fewer than 4 days a week). Long run no more than 30–35% of the week below 40 km a week, 25–30% above 60 km. | D (convention) |
| VL-7 | Last 8–12 weeks: marathon-pace sections in the last 20–40% of some long runs, alternating with easy long runs. | D (elite practice) |
| VL-8 | Progression option: first 60–70% easy, next 20–30% building to marathon pace, last ~10% at marathon to half-marathon effort. At most every 2–3 weeks. | D |
As an example of the percentage caps: at 40 km a week, 35% is 14 km. At 60 km a week, 25–30% is 15–18 km. And at 6:00 per km, a 3-hour cap means about 30 km.
The spike rule always wins. Whatever these caps allow, no run should be more than 10% longer than your longest run of the past 30 days. That rule has the best injury evidence. Read the long-run injury rule, and see why the weekly 10% rule doesn’t hold up.
What we don’t know
- No trial has tested long-run length against marathon results. All the volume evidence is observational.
- Durability findings are recent and small in some cases, and whether durability can be trained is unresolved.
- Every specific cap is untested, including the ones in the most popular plan books.
- The sex difference in fatigue comes from one study and needs confirming.
Sources
- Doherty et al. (2020). Meta-regression of training volume and marathon performance.
- Tanda and Knechtle (2013). Model of training volume and marathon time in recreational runners.
- Toresdahl et al. (2020). Training and injury in first-time New York marathoners.
- Maunder et al. (2021). Durability as a predictor of endurance performance.
- Jones et al. (2024). Durability and marathon performance.
- Smyth et al. (2022). Heart-rate and pace decoupling in 82,303 marathoners.
- Hunter and Muniz-Pumares (2025). Lactate-threshold decline after a 90-minute run in London finishers.
- Zanini et al. (2026). Economy loss during 90-minute runs by long-run habit.
- Jaén-Carrillo et al. (2026). Sex differences in speed loss after prolonged running.
- Jones and Kirby (2025). Expert guidance on training durability.
- Hunter (2026). Whether durability is trainable or a trait.
- Haugen et al. (2022). Review of how world-class distance runners train.
Last reviewed 13 November 2026.