The short answer: among 197,643 runners who ran a second marathon one to three years after their first, 56% ran faster the second time, and the median change was 2.9 minutes faster (about 1.2% of the first time). The typical second marathon is not a breakthrough. It is a modest step: a 4:00 to 4:30 first-timer gained 3.2 minutes on median, while a sub-3 first-timer was slightly slower on median. The big gains belong to the slowest debuts: people who first finished in over 6 hours knocked off 15 minutes on median, and two in three of them improved.
Those numbers come from the official results of the New York City, Berlin and Chicago marathons, where the same person can be matched across editions. The charts below show how the change depends on your first time, how spread out the outcomes are, and what happens over a fifth marathon.
What to expect from your second marathon
The clearest pattern is that the gain depends almost entirely on how fast the first one was.
| First marathon | Runners | Ran faster | Median change | Middle half of runners | Improved by 10+ min |
|---|---|---|---|---|---|
| Sub-3:00 | 6,306 | 45.5% | +1.0 min | -4.2 to +9.2 min | 8.2% |
| 3:00 to 3:30 | 21,341 | 50.0% | 0.0 min | -7.8 to +11.7 min | 19.4% |
| 3:30 to 4:00 | 43,792 | 52.5% | -1.0 min | -11.1 to +12.3 min | 27.4% |
| 4:00 to 4:30 | 45,763 | 56.1% | -3.2 min | -16.2 to +12.9 min | 36.5% |
| 4:30 to 5:00 | 38,213 | 58.2% | -5.1 min | -21.8 to +12.9 min | 41.8% |
| 5:00 to 6:00 | 34,134 | 61.1% | -8.6 min | -29.7 to +12.5 min | 48.0% |
| 6:00+ | 8,094 | 67.3% | -15.2 min | -40.7 to +8.2 min | 56.3% |
Negative numbers mean a faster second marathon. In percentage terms the median gain runs from 0.6% slower for the sub-3 group to 4.0% faster for the 6-hour group.
Part of this is genuine improvement: a first marathon is usually run on the least training and the least experience a runner will ever have, and the slower the debut, the more room there is. But part of it is simple regression to the mean. A 5:30 first marathon is more likely to have been a bad day (heat, cramps, a walk from 30 km) than a 3:10 first marathon is, and a bad day tends to be followed by a normal one. A sub-3 debut, on the other hand, was probably a good day, so the next race is more likely to be ordinary. The data cannot separate the two effects, so treat the bucket medians as "what typically happens next", not as "what training will give you".
How spread out the outcomes are
The median hides how varied second marathons are. More than a quarter of runners finished at least 10 minutes slower.
Across all 197,643 pairs:
- 56.0% ran faster, 44.0% ran slower.
- 35.6% improved by 10 minutes or more.
- 27.8% got slower by 10 minutes or more.
- The middle half of runners landed between 16.5 minutes faster and 12.2 minutes slower.
- 2.3% improved by more than an hour and 2.5% were more than an hour slower.
So the realistic range for a second marathon is wide. Roughly one runner in three makes the kind of double-digit gain that training plans promise, one in three lands within 10 minutes of the first time either way, and one in four to one in three goes backwards by 10 minutes or more.
Does the gap between marathons matter?
Yes, and sooner is better in this sample.
| Years between marathons | Pairs | Ran faster | Median change |
|---|---|---|---|
| 1 year | 112,214 | 58.4% | -3.7 min |
| 2 years | 56,140 | 54.4% | -2.3 min |
| 3 years | 29,289 | 49.8% | +0.1 min |
Runners who came back the following year improved on median; runners who waited three years were a coin flip. That is consistent with fitness carrying over from one build to the next, and with the three-year group being on average older and more likely to have had a break.
Men and women improve by the same amount: 56.0% of men ran faster (median 2.8 minutes) and 55.9% of women (median 3.0 minutes). Running the same race again versus a different one of the three made no difference either: 56.1% improved on a repeat course, 55.8% on a new one.
What happens over five marathons
For the 20,252 runners in the sample with at least five verified marathons, the story splits in two.
| Marathon | Median time versus the first | Median personal best so far | Faster than the first |
|---|---|---|---|
| 1st | 0 | 0 | |
| 2nd | -3.6 min | -3.6 min | 57.6% |
| 3rd | -3.5 min | -10.2 min | 56.0% |
| 4th | -1.3 min | -13.3 min | 51.9% |
| 5th | +3.7 min | -15.0 min | 45.1% |
The median personal best keeps falling: 10 minutes under the debut by the third marathon, 15 minutes under by the fifth. But the median time of the fifth marathon itself is 3.7 minutes slower than the first, and fewer than half of fifth marathons beat the debut. These runners are typically five to eight years older by their fifth race, and the age effect eventually outweighs experience. The practical reading: most experienced marathoners set their best in their second to fourth race, and later marathons are about holding on rather than improving.
Where the data comes from
The results are from the Pace Notes open dataset (CC BY 4.0), which compiles official finisher results and links the same runner across editions by name, sex and age pattern. We used verified marathon editions of the New York City Marathon (2006 to 2026), Berlin Marathon (2005 to 2026) and Chicago Marathon (2005 to 2026), kept only high and medium confidence person links, and took each runner's first two marathons in the dataset where the second came one to three years after the first. Pairs with a first marathon faster than 2:30 or slower than 7:00 were dropped. That left 197,643 pairs: 121,552 men and 76,091 women. The median first marathon was 4:18:11 and the median second was 4:14:45.
Two caveats matter. First, "first marathon" means first in these three races. Someone whose actual debut was a smaller race elsewhere counts here from their first New York, Berlin or Chicago. Second, the sample only contains people who came back. Runners who had a miserable debut and never ran another marathon are invisible, which makes the improvement figures look a little better than they would for every first-timer. The direction of every finding holds, but the exact shares should be read with that in mind.
Frequently asked questions
How much faster is a second marathon on average?
The median improvement across 197,643 runners was 2.9 minutes, or about 1.2%. The mean was 1.7 minutes because the slow tail of bad second races pulls it up. For a 4:00 to 4:30 first-timer the median gain was 3.2 minutes; for a 5:00 to 6:00 first-timer it was 8.6 minutes.
What percentage of runners improve on their second marathon?
56%. That rises to 58% when the second race is the following year and falls to 50% when it is three years later. Among sub-3 first-timers only 45% improve, because a sub-3 debut is usually already a good day.
Is it realistic to take 15 or 30 minutes off?
It depends on where you start. 36.5% of 4:00 to 4:30 first-timers improved by 10 minutes or more, and a quarter of them improved by more than 16 minutes. A 30-minute gain was common only for debuts over 5 hours: a quarter of the 5:00 to 6:00 group improved by 29.7 minutes or more.
Why do fast first-timers get slower?
Regression to the mean. A sub-3 or 3:00 to 3:30 debut usually means the runner's preparation and race day both went well, so the next attempt is more likely to be ordinary than better. The median change for sub-3 debuts was 1.0 minute slower, with half of them landing between 4 minutes faster and 9 minutes slower.
Should I run my second marathon next year or wait?
The data favours next year. Runners who returned after one year improved on median by 3.7 minutes and 58% ran faster; after three years the median change was zero. Fitness carries over, and a year is enough to fix what went wrong without losing the base. See the taper guide for the final week and the pacing guide for the most common reason second marathons go wrong.