Estimating Your NYC Marathon Finish Time

A marathon goal is a prediction. The inputs are a recent race, the training that can support it, and a course that is not a track. New York is none of the flat equivalents people type into a calculator.

Today is race day here. If you are lining up this morning, or doing the math for a future November, here is how I would estimate a finish time and then spend it. I start with a recent raced half, 25k, or 30k — not a training loop at "race pace."

Riegel's formula, and why it runs hot

Peter Riegel's 1977 rule fits on a napkin:

T2 = T1 × (D2 / D1)^1.06

T1 is the time you have. D1 is that distance. D2 is the marathon. The 1.06 exponent is a generalization across distances, not a law of physiology. The marathon is almost exactly twice a half, so the ratio is 2. 2^1.06 is about 2.085. A 1:45:00 half is 105 minutes: 105 × 2.085 ≈ 218.9 minutes → 3:38:55.

That is a flat-course equivalent for a runner trained for both distances. It tends to be optimistic for the marathon. The exponent cannot see glycogen past the window a half never enters, whether you practiced 18 to 22 mile long runs, or heat, hills, wind, and a crowded first 5k.

I treat Riegel as a ceiling. If the napkin says 3:38, I do not line up at 3:35. Some coaches use a steeper exponent, around 1.08 to 1.10, as a second napkin. The honest goal usually lives closer to the slower number unless the training is obviously marathon-specific.

VDOT and the Daniels tables

Jack Daniels' VDOT tables do the same job with more empirical smoothing. You look up a recent race, read an index, then read across to the marathon column and to training paces. I use them as a well-trained equivalent, not a weather-and-course forecast. They assume you are trained for the distance you are predicting. A VDOT from a strong half does not automatically transfer if your longest run is 14 miles.

The training paces are often the more useful output. If they are slower than anything you have been willing to run on Sundays, the table is describing a runner you have not built yet. When Riegel and Daniels disagree, I take the slower equivalent, then I apply the checks below.

Adjust for long runs and weekly mileage

The last 10k is a skill. If the longest run in the block is well short of 18 to 20 miles, I would not take a raw equivalent at face value. A 20-miler closed in control supports a goal near the formula. A 20-miler you faded says the formula is describing a better-fueled version of you.

Weekly mileage is the other check. A half raced off a thin week is often sharper than the marathon that week can support. Has the volume been consistent enough that 26.2 is a longer day, not a different sport? If the block was only easy miles and one hard half, I trust the half as fitness and distrust it as a marathon prediction. I move the goal in minutes, not fake percentages.

Why New York is slower than a flat course

The race crosses five bridges: the Verrazzano-Narrows at the start, the Pulaski from Brooklyn into Queens at about halfway, the Queensboro from Queens into Manhattan around miles 15 to 16, the Willis Avenue into the Bronx around mile 20, and the Madison Avenue back into Manhattan shortly after. Each one is a climb, a descent, a wind corridor, or all three.

The Verrazzano opens uphill onto the span — the high point of the course — then drops into Brooklyn. The downhill is where people trash their quads before the boroughs have started. Brooklyn is the longest stretch and the easiest place to believe you are fitter than you are.

The Queensboro around 15 to 16 is the first real late-race hill, and quieter on the span. Then you dump onto First Avenue: wide, downhill at first, packed with noise. A hard bridge plus a runway of spectators is how people run their fastest 5k of the day at mile 17. The Bronx bridges are shorter and poorly timed. They arrive when you are already on tired legs.

Fifth Avenue around miles 23 to 24 is a long grind before the park. It is not steep like the Verrazzano. It is late. Central Park after that is rollers, not a cooldown.

Then the non-elevation taxes. NYC starts in waves and corrals, so early miles can be a shuffle. The first Sunday in November can be cool, cold, or unseasonably warm. The bridges are exposed. Wind there is a real cost. Heat is the variable I respect most.

A rough rule of thumb: I add a couple of minutes versus the same fitness on a flat, uncrowded course, and more than that if it is warm, windy, or I expect a dense first 8k.

A worked example

Take a recent half in 1:45:00. Honest effort, inside the training block.

Riegel: 105 × 2.085 ≈ 3:38:55. Call it 3:39. A Daniels lookup should sit in the same neighborhood; if it is slower, keep the slower number.

Suppose long runs reached 18 to 20 and finished in control, and weekly mileage was consistent. I would not throw the equivalent out. I also would not take 3:39 as the race plan. Add a small NYC buffer — a couple of minutes, more if the forecast is warm.

I would write the goal as 3:45, not 3:39: Riegel's known optimism plus a course that is not Berlin. Stronger training — goal-pace finishes on long runs, a recent 30k that already looks like 3:40 — might get me nearer 3:42. I would still not start at 3:38 pace and "see how I feel."

3:45 is 8:35 per mile on a 26.21875-mile course. That is the wrist-card number. Not 8:21 (3:39), and not 8:10 (the Verrazzano downhill lying to you). Reuse it: half in minutes × 2.085, convert back, then add a training holdback and a small NYC buffer.

Mile paces for common goals

Even-pace numbers for 42.195 km, rounded to the nearest second.

Goal	Pace / mile
3:00	6:52
3:15	7:26
3:30	8:01
3:45	8:35
4:00	9:09
4:15	9:44
4:30	10:18
5:00	11:27

If your watch is already 20 seconds fast at mile 3, you are spending that table, not banking it.

How I would pace the day

Do not bank time on the Verrazzano. The first mile will be slower than goal on the clock. That is correct. The second mile will tempt you to take it back plus interest. The downhill is eccentric load you need at mile 23.

Run the bridges by effort. On the Queensboro, Willis, Madison, and Fifth Avenue, same effort as flat goal pace looks slower going up and faster coming down. Chase the number and you spike on the descents.

Aim even-to-negative, knowing NYC fights you. A true negative split is hard because the late course is the hilly part. I want a first half slightly slower than goal, a controlled First Avenue, and enough left that Fifth Avenue and the park are still running. If the first 10k is your fastest, the prediction is already wrong. Through Brooklyn, if it feels easy, you are probably on pace.

Fuel like the estimate depends on it. The formula assumed you would still have glycogen in a window a half never sees.

Trust mile markers more than GPS

Manhattan is a canyon. The bridges are steel and water. Watches add distance among tall buildings, wander on the spans, and will tell you that you have already run 26.4 with the finish still ahead. That drift makes people surge: the watch says 8:50, the effort is fine, you "correct," and you are racing a phantom.

Pace off official mile markers and elapsed time, plus effort on the bridges and in the park. If the watch says you are behind and the markers say you are on, the markers win.

A checklist I would reuse

• Recent raced effort, not a hopeful training loop
• Riegel (× 2.085 from the half) written down as a ceiling
• Daniels / VDOT equivalent checked; keep the slower of the two
• Long runs and weekly mileage actually support 26.2, or the goal moves
• NYC buffer added: a couple of minutes as a baseline, more for heat, wind, or a crowded wave
• Goal pace on a card, plus a "do not go faster than" first 10k
• Effort on the five bridges, not splits; no banking on the Verrazzano downhill
• First Avenue treated as a hazard; something left for Fifth Avenue and the park
• Forecast checked the morning of; warm means the number gets slower
• Watch is a backup; mile markers and effort are the race

If any of those are missing, I have a wish.

The number is a budget

Write the equivalent from a real race. Discount it for the training you actually did. Discount it again for New York. Then run the first 10k like someone who believes the last 10k exists. Riegel will give you a flattering time. The Verrazzano will offer to cash it in before Brooklyn starts. I would keep the receipt until the park.