Does a Bigger Field At the NYC Marathon Mean Slower Times?

Feature photo: Martineric from Lille, France., CC BY-SA 2.0, via Wikimedia Commons

In the lead up to the 2025 New York City Marathon, the Economist published a piece about the race. A central claim of the article, based on an analysis of finish times going back to the 1980’s: as the race fills with amateurs, speed matters less than spirit.

In short, their main thesis was that as interest in the race increased and the field grew, the slower runners were getting increasingly slow. This claim reminded me a lot of the RunRepeat “mega study” from a couple years ago that argued that marathon runners in general were getting slower. That led me down a giant rabbit hole, and my main conclusion was that what they had really “discovered” was that the average age of marathon runners was increasing.

The Main Claim

Let’s start with what the main claim of the article is:

An analysis of every finisher’s time shows that, for men and women alike, the leading 10% of runners have maintained their pace. Top men typically finish in just over three hours, while the top women are 30 minutes slower. But the back of the pack has steadily fallen behind. In 1986 the slowest tenth of men took just under five hours to finish the race, while women took five and a half. By 2022 each group slowed by roughly an hour.

Generally, they’re saying that the fastest runners have stayed the same while the slowest runners have gotten significantly (about an hour) slower. They’ve also marked out a clear timeframe for comparison – 1986 to 2022.

By “fastest” runners, they specifically mean the runner at the 10th percentile. If there were 100 runners, this would be the 10th finisher. By the “slowest” runners, they specifically mean the runner at the 90th percentile. If there were 100 runners, this would be the 90th finisher.

Using that same methodology, I prepared my own version of their chart below.

At first glance, their claim seems fairly accurate. If you focus on the bottom of the visual – the fastest runners – the finish times are pretty stable. They’re a little faster in the 80’s and a little slower from the 1990’s through the 2010’s. But there’s no big or obvious shift.

In the top half of the visual, however, both lines move up and to the right pretty clearly. There’s some bouncing up and down year to year, but a straight line fit through those dots would certainly show an increase in finish times. And the net difference from 1986 to 2022 is about an hour.

The Main Problem

There might be some truth to this claim, but there’s at least one big problem with it. This data is for all runners, and besides the fact that there are more runners today than there were forty years ago, something else has changed. Runners are, on average, older.

The first large generation of marathoners picked up the sport in the 1970’s and the 1980’s, thanks in large part to the way that the New York City Marathon turned the race into a mass participation event. At the time, the field trended much younger. But over time, some of those original runners stuck around and filled up the ranks of Masters runners.

I previously did a deep dive on the changing demographics of runners at the New York City Marathon here, but the visual below summarizes the data in the context of our earlier assumptions. Specifically, it shows how the 10th and 90th percentile of runner’s ages have changed.

The bottom lines show how young the youngest portion of field is. For women, that 10th percentile is between 25 and 27 years old, and for men it’s between 27 and 29. From 1986 until 2025, it’s pretty constant.

The top lines show how old the oldest part of the field is. At the 90th percentile, women have gone from 46 up to 55 and men have gone from 50 up to 58. The oldest ten percent of the field is now about a decade older than it was back in the 1980’s, and there are significantly more runners in their 60’s and 70’s.

This introduces a huge confounding factor into the mix. Older runners tend, on average, to be slower than younger runners. So while the youngest part of the field is still in their 20’s and presumably running similar times, the oldest part of the field is drifting upwards in age. There’s not a huge performance difference between runners in their 30’s and 40’s, but it becomes increasingly stark for most runners – especially the ones that aren’t heavily trained – as they hit their 50’s and 60’s.

Here’s another version of the original chart. But here, there’s one line for all runners and one just for runners under 40. The dropdown box toggles between men and women.

When you focus just on runners under 40, there is still an upward shift in the finish times for slower runners. But the net change from 1986 to 2022 is smaller, and you can see over time how the line for all runners drifts up and away from the line for runners under 40.

What If We Hold Age Constant?

Let’s take another look at the original visual, but this time let’s hold age constant over the time period.

This version of the visual below shows the 10th and 90th percentile among male finishers at the New York City Marathon, but it restricts results to the same age group. The dropdown menu allows you to switch between different age groups.

For men under 40, the line is pretty flat from the mid 1990’s to the mid 2010’s, and then it’s flat again from 2014 to 2019. The results are fairly similar for men in their 40’s, although the gap between 2013 and 2014 starts to fade in the older age groups.

Here’s the same visual for women.

Again, for women under 40, you’ve got a slight increase in the early years, stagnation in the middle, and a jump to a new plateau in the later years. For women in their 50’s, the slowest times increase and then drop suddenly from 2006 to 2019. Same thing for runners in their 60’s.

So it’s not as simple as – more people are running the New York City Marathon, and the slowest runners are finishing slower. Age plays a big role in that equation, and something else seems to be happening between 2013 and 2014.

Where Are the the Outlier Years?

The trend is also obscured a bit by some outliers. To help better identify the trend, I’ve labeled several outlier years on the graph below.

The first outlier, 1990, is labeled because it was warmer than usual. The high temperature in the early afternoon was in the 70’s – not extremely hot, but warmer than usual. And certainly warm enough to have an impact on runners who are out there in the heat of the day (i.e. the slowest runners).

From 2003 to 2005, it was also warmer than usual. In 2003, the maximum temperature wasn’t too high, but overnight lows the night before were warm. It was in the 60’s at the start, and it rose to the high 60’s throughout the afternoon. Conditions were similar in 2004 and 2005. These years were warm enough to increase finish times for even the fastest runners.

2001 is another strange year. That was the year of 9/11. Although the race took place, the field was much smaller than usual. There’s a little dip in the finish times, and it’s plausible that fewer casual runners took part – leaving a relatively stronger field in that particular race.

Other than the cancellations (2012 and 2020), the other marking here is 2022. Throughout this entire period, this was probably the worst weather in the history of the race. There were a few hot and humid days in the 1970’s and early 1980’s, when the race still took place in late October. But since the race moved to November in 1986, there hasn’t been a day this warm and humid. The temperatures were already in the 70’s with a dew point in the 60’s at the start, and it remained muggy and warm all day.

The finish times in 2022 were definitely slower than usual.

So What’s Actually Happening Here?

If you set aside those outlier years for a moment (1990, 2001, 2003-05, 2022), the story of the graph looks a little different.

From 1986 until around 2000, the finish times of the slowest men and women do steadily (but slightly) increase by about 20-25 minutes. But from about 2000 / 2002 through 2013, they’re actually quite stable – other than the outlier years. The slowest women are pretty consistently between 5:40 and 5:45, while the slowest men are between 5:10 and 5:15.

Then, there’s a marked increase from 2013 to 2014. Other than 2022, the remaining period of 2014 through 2025 is fairly stable as well. In that last period, the slowest women hover around 5:55 to 6:00, while the men are around 5:25 to 5:30.

There’s not a sustained increase over the full time period, and the net change from the late 80’s to the early 2020’s is more like 40-45 minutes as opposed to an hour.

The real interesting thing to me, though, is that there’s a pretty marked difference between 2013 and 2014. That kind of jump, with relative plateaus on either side, suggests that something happened that year.

At first I thought maybe the wave start times and the finish line closing time might have changed. But that doesn’t appear to be the case. Then, I remembered that the rules for guaranteed entry changed pretty significantly in 2014.

There are two specific changes that I think could have had an effect specifically on the slowest part of the field:

1. Runners could no longer defer their entry indefinitely. They can defer for one year, and then they lose their spot.
2. Runners who missed the lottery in three consecutive years were no longer guaranteed entry.

Previously, if you earned entry into the race, you could continue to defer that entry for multiple years. Perhaps you got injured or life got in the way and you didn’t train. Now, you could only defer once – and then you had to use it or lose it.

The logical impact of this change is that some number of runners will likely choose to run the race when they aren’t prepared because they no longer have an option to defer. That’s more likely to add slower runners to the field than faster runners.

With the lottery, you were previously guaranteed a place if you missed out on the lottery three times in a row. But starting with 2014, that would no longer be the case. The old system gave a boost to people who were consistently interested in running the marathon and repeatedly entered the lottery. Now, that repeat entrant has an equal chance of running the race as the first time entrant.

The logical impact of this change is that the pool of runners will shift towards first time lottery applicants. And at least some of these runners – likely more than the repeat entrants – are poorly trained.

What’s the Bottom Line?

For one final visual, let’s add a third group into the mix – the median finish time.

When you look at the top of the visual, the finish time among the slowest runners has been slower over the past 10 years than it was previously. But the median finish times and the fastest finish times have both stayed pretty consistent, at least through the 1990’s.

When you hold age constant – and the visual above only includes runners under 40 – there’s a gradual increase in finish times for the first fifteen years or so. That increase is starker among the slowest runners, and it’s more subtle among the median and faster runners.

But after 2000, aside from the outliers, things are pretty stable with one exception – the shift that happens between 2013 and 2014. There’s a big increase for the slower runners and a much smaller one for the average runners.

The fact that there’s a pretty significant change from one year to the next suggests that it’s a result of something else that changed. And my best guess is that it can attributed to the change in registration rules beginning with the 2014 race.

The initial trend, identified in the Economist article, is at least in part driven by the aging of runners. By holding age constant and looking at the trends within age groups, it’s easier to isolate that 2013-2014 change and see that it’s significant.