How Many Runners Negative Split the 2025 New York City Marathon?

Feature photo by Marc. A Hermann of the MTA - CC BY 2.0

The New York City Marathon is run on what is typically considered the toughest course of the original six Majors. Between the bridges and the hills into Central Park, there’s quite a bit of elevation gain throughout the race, and it doesn’t benefit from the big drop that Boston has early on.

The race results included detailed splits at every mile for every runner. So I thought it would be interesting to dig into this data and see where runners slowed down – and how many blew up later in the race. My deep dive on the 2025 Chicago Marathon also included an examination of runners’ splits, so this is a good opportunity to compare the two races.

How Many Runners Negative Split the 2025 NYC Marathon?

Let’s start with some top line numbers. Across the board, 6.3% of men and 7.9% of women ran a negative split at the 2025 New York City Marathon. This means their second half was faster than their first half.

Right off the bat, here’s an indication that New York is a tougher course than Chicago. At that race, 10.7% of men and 12.2% of women – almost twice as many – ran a negative split. The hills at New York, especially the ones in the second half, make it tough to negative split. However, one thing does remain the same: women are more likely to negative split than men.

Here’s another topline stat for New York. Among men, 17.4% of finishers ran a large negative split where their second half was at least 25% slower than their first half. For context, that means if a runner started out with a 90 minute first half, they finished up with a 112.5 minute second half – for an overall finish time of 3:22:30. That’s a pretty big blow up.

By contrast, only 7.7% of women had a second half where they slowed down by at least 25%. Not only are women more likely to run negative splits, they are also more likely to run a moderate positive split – and much less likely to completely blow up.

The visual above breaks this data out by how fast a runner started the race. Along the x-axis, runners are grouped into 5 minute increments. The 1:30 mark, for example, represents anyone who started the first half with a time of 1:30:00 to 1:34:59.

The darker blue at the bottom is the small share of runners who ran a negative split. Faster runners tended to be slightly more likely to negative split than slower runners.

The light blue above that represents runners who finished slightly slower – 0-10% slower in the second half. Among the fastest runners, about 2/3 fit into this bucket. But by the time you get to moderately fast paces (1:30-1:45 for men, 1:40-1:55 for women), that’s dropped off quite a bit. And it continues to drop from there. But the low point – after around the 2:00 mark – is much lower for men than it is for women.

The other big difference is in the pink group at the top – runners who were 25-50% slower. Relatively few women fall into this bucket. Even at the slow end of things, only 10% of women have this dramatic of a blow up. But among the slower men, anywhere from 20-25% of men had a dramatic blow up.

The overall pattern here – that men are more likely to run positive splits, and those splits become bigger among slower men – is similar to the overall pattern from Chicago. The main difference is just that more runners were able to negative split in Chicago than in New York.

Which Splits Were Slowest at the 2025 NYC Marathon?

The individual split data included the runner’s time at the three mile mark and then every mile there after. The 3M split below represents the average split for the first three miles, and every mark thereafter is for the pace in a given mile. I discarded the final 0.2 miles at the end of the race for this visual.

In some cases, split data wasn’t available for an individual runner for an individual mile. In these cases, I averaged the difference of the next available splits.

Although the three biggest hills in the race are the Verrazano Bridge, the Queensboro Bridge, and the final stretch into Central Park, there are smaller rollers throughout the course. So I was curious to how this lines up with where runners slowed down the most.

This visual identifies the slowest split throughout the course for each runner and counts up the total number of runners.

There’s one split that stands out – Mile 24. This is the climb up Fifth Avenue on your way into Central Park. Almost 20,000 finishers – about a third of the field – had their slowest split at this point.

The other two big outliers are Mile 15 and Mile 20.

Mile 15 is the Queensboro Bridge. About five percent of runners slowed down the most at this point – not a lot in the scheme of things, but a lot more than the surrounding miles.

Mile 20 is the Willis Ave Bridge. Over 10% of runners had their worst split here, although there are quite a few runners who hit their slowest split anywhere form Mile 19 to Mile 26.

So generally speaking, runners struggle with Fifth Ave, the Willis Ave Bridge, and the Queensboro Bridge, and they generally slow down towards the end of the race.

Which Splits Were Fastest At the 2025 NYC Marathon?

On the flip side, which of these splits were runners’ fastest? And is there any pattern behind this data?

A lot of runners – the vast majority – had their fastest split during the first quarter of the race. From Mile 3 through Mile 7, many runners lay down their fastest split – and they’ll eventually slow down from there.

One of the biggest outliers is Mile 10. This mile is pretty much all downhill after a short hill in Mile 9. A decent number of runners – about 6-7% – ride this decline to their fastest split of the race.

There’s also a spike at Mile 18. At this point, you’ve made it up and down the Queensboro Bridge, and you’ve got a little hill to run down. You might even be thinking about the finish – and how you feel great. But those hills at the end are waiting.

Then, there’s an interesting spike at Mile 26. It’s normal to push the pace at the very end of the race – but unless you’re feeling great that’s often in the last few hundred meters, not for a full mile.

I took a closer look at this group, and it’s about 60% women. Among the men, the biggest group started out at around 1:55-2:00 for the first half – so on pace for a 4:00 finish. Among the women, there were big clusters around 1:55-2:00 and 2:10-2:15. A lot of these runners are people likely targeted a big round number (4:00 / 4:30), and for this group it was probably a conservative estimate that left a lot of gas in the tank.

How Much Do Splits Deviate From the Average?

The previous visuals just looked at the extremes, so it doesn’t give you a good idea of how things varied along the whole course. Just because a runner was the slowest at Mile 24, it’s hard to say how much slower they were than at surrounding miles.

The visual below looks across the entire course and calculates how much the pace at that mile varied from the runner’s average pace for the entire course. A negative number means they were faster at that spot and a positive number means they were slower.

The data is broken out by Men (top) and Women (bottom), although the pattern is the same.

Runners tend to be faster in the first half of the race. The first seven miles or so are the fastest, and then they gradually slow down through the halfway point. There’s a spike at the Queensboro Bridge (Mile 15), followed by a few better miles.

Then, from Mile 19 to Mile 26, people tend to be a lot slower. This matches up with the overall data that it’s rare for runners to run a negative split – which would require them to be faster in the second half.

Between men and women, the big difference is in magnitude. On the left, faster end of the things, men tend to be further below their average pace than women. But on the flip side, they tend to slow down more in the later part of the race. In other words, women tend to run a more even race.

Conclusion

Here’s one final fun fact. You probably read that Chelsea Clinton ran the New York City Marathon this year, and you may have read that she had impressively even splits. She did – although interestingly, her slowest splits were Mile 15 (+25 seconds) and Mile 24 (+18 seconds), and her fastest split was Mile 26 (-33 seconds).

I was curious how many runners had more even splits – defined here by the maximum deviation from the average pace. Chelsea’s biggest difference was Mile 26 (-33 seconds), and 4,241 other runners ran a race that deviated less than 33 seconds from their average pace. If you ignore Mile 26, when she sped up, and instead base it on her slowest split (-25 at Mile 25), there are 1,939 other runners who deviated that much or less.

The winner for the Most Even Split Award? Edward Katz. The man ran like a metronome. Through Mile 16, he ran every mile between 6:30 and 6:37. He picked up the pace slightly from there and finished the remainder of the race between 6:21 and 6:30.

If you have any other questions about the splits at the New York City Marathon, leave a comment below – and I’ll see if I can come back to the topic later and look into it.