Race time predictor
Predict your 5K, 10K, half marathon and marathon from one recent result — with two models side by side, so you can see how wide the prediction really is.
Predict your time at another distance
Enter a recent race result. Both models are shown because they disagree, and the gap between them is the honest width of the prediction.
- Your VDOT
- 38.3
- 5K pace
- 5:00
- Seed time
- 25:00
Daniels equivalent
per kilometre
5K
| Distance | Riegel | Daniels | Pace /km |
|---|---|---|---|
| 1 mile | 7:31 | 7:24 | 4:38 |
| 10K | 52:07 | 51:53 | 5:12 |
| 10 mile | 1:26:19 | 1:26:11 | 5:22 |
| Half marathon | 1:55:00 | 1:55:04 | 5:27 |
| Marathon | 3:59:47 | 3:57:55 | 5:40 |
Both models assume you are as well trained for the target distance as for the one you raced. Neither knows how many long runs you have done, so both will flatter a marathon predicted off a 5K.
Why two models disagree
A race predictor is a claim about how fast you slow down as the distance grows. There is more than one defensible way to make that claim, so this page shows two.
Riegel
Pete Riegel's endurance formula is one line: T2 = T1 x (D2/D1)^1.06. He fitted the 1.06 exponent across world records from 1500 m to the marathon. It is simple, transparent, and you can check it on a calculator.
Daniels
The VDOT approach goes through physiology instead. It converts your result into a fitness score using the oxygen cost of the speed you held and the fraction of maximum sustainable for that duration, then reads the other distances off the same score.
Which is right
Neither, exactly. Over short extrapolations they agree closely — from a 25:00 5K, Riegel says 52:07 for 10K and Daniels says 51:53. Over a marathon the gap opens: 3:59:47 against 3:57:55. The size of the disagreement is the useful output. If two independent models built from different data land two minutes apart, the honest prediction is a range, not a time.
Equivalent times from a 5K
Riegel equivalents for every 5K time from 15:00 to 40:00.
| 5K | 1 mile |
|---|---|
| 15:00 | 4:31 |
| 16:00 | 4:49 |
| 17:00 | 5:07 |
| 18:00 | 5:25 |
| 19:00 | 5:43 |
| 20:00 | 6:01 |
| 21:00 | 6:19 |
| 22:00 | 6:37 |
| 23:00 | 6:55 |
| 24:00 | 7:13 |
| 25:00 | 7:31 |
| 26:00 | 7:49 |
| 27:00 | 8:07 |
| 28:00 | 8:25 |
| 29:00 | 8:43 |
| 30:00 | 9:01 |
| 31:00 | 9:19 |
| 32:00 | 9:37 |
| 33:00 | 9:55 |
| 34:00 | 10:13 |
| 35:00 | 10:31 |
| 36:00 | 10:50 |
| 37:00 | 11:08 |
| 38:00 | 11:26 |
| 39:00 | 11:44 |
| 40:00 | 12:02 |
Riegel’s endurance formula: T₂ = T₁ × (D₂/D₁)1.06, fitted across race records from 1.5 km to the marathon. It assumes you are equally trained for both distances, which is the one assumption a runner predicting their first marathon off a 5K is breaking.
Equivalent times from a 10K
A 10K is a better seed than a 5K for anything half marathon or longer — it is closer to the target and more aerobic.
| 10K | 1 mile |
|---|---|
| 32:00 | 4:37 |
| 34:00 | 4:54 |
| 36:00 | 5:12 |
| 38:00 | 5:29 |
| 40:00 | 5:46 |
| 42:00 | 6:03 |
| 44:00 | 6:21 |
| 46:00 | 6:38 |
| 48:00 | 6:55 |
| 50:00 | 7:13 |
| 52:00 | 7:30 |
| 54:00 | 7:47 |
| 56:00 | 8:05 |
| 58:00 | 8:22 |
| 1:00:00 | 8:39 |
| 1:02:00 | 8:57 |
| 1:04:00 | 9:14 |
| 1:06:00 | 9:31 |
| 1:08:00 | 9:48 |
| 1:10:00 | 10:06 |
| 1:12:00 | 10:23 |
| 1:14:00 | 10:40 |
| 1:16:00 | 10:58 |
| 1:18:00 | 11:15 |
| 1:20:00 | 11:32 |
Riegel’s endurance formula: T₂ = T₁ × (D₂/D₁)1.06, fitted across race records from 1.5 km to the marathon. It assumes you are equally trained for both distances, which is the one assumption a runner predicting their first marathon off a 5K is breaking.
The assumption both models break
Riegel and Daniels both assume you are equally trained for both distances. That assumption is fine when you predict a 10K from a 5K, and it is close to false when you predict a marathon from a 5K.
A marathon has a specific failure mode neither model can see: running out of glycogen, and the muscular damage of three or four hours of impact. Those are trained by long runs, not by fitness in the abstract. Two runners with identical 5K times, one with a 32 km long run behind them and one with a 14 km long run, will finish a marathon twenty minutes apart, and the predictor will give them the same number.
Practical rule: trust a prediction across one step of distance, discount it across two, and treat three as entertainment. A marathon predicted from a 5K is three steps.
How to get a prediction worth having
- Seed from the closest distance you have raced. Predicting a marathon? Use a half, not a 5K. The error compounds with every step.
- Use a real race, not a solo time trial. A time trial is typically 1–3% slower than a race over the same distance, because pacing alone is harder than pacing in a field.
- Use a recent one. Six weeks is fresh, six months is a historical document.
- Adjust for the day. Heat above roughly 15°C, wind, and elevation all cost time that no exponent knows about.
Once you have a number you believe, the pace chart for that distance turns it into splits you can actually run to.
Close the gap between prediction and result
7-day free trial · then $12.99/mo · cancel anytime
Fair questions
How accurate is a race time predictor?
Accurate across one step of distance, optimistic across three. From a 25:00 5K the two models here predict a 10K within 14 seconds of each other (52:07 and 51:53), but a marathon nearly two minutes apart (3:59:47 and 3:57:55) — and both assume marathon-specific training that a 5K result cannot demonstrate.
What marathon time does a 25-minute 5K predict?
3:59:47 by Riegel, 3:57:55 by Daniels — so, about four hours, if you have done marathon training. Without long runs, treat it as a ceiling rather than a forecast.
What is the Riegel formula?
T₂ = T₁ × (D₂/D₁)^1.06, from Pete Riegel, first published in Runner's World in 1977 and set out in full in American Scientist 69(3) in 1981. The 1.06 exponent was fitted to race records between 1500 m and the marathon, so the formula is at its best inside that range and increasingly optimistic outside it.
What half marathon time does a 50-minute 10K predict?
1:50:19 by Riegel. Daniels puts a 50:00 10K at VDOT 40.0, which predicts 1:50:52 — a 33-second spread, which is about as tight as two independent models get.
Should I use a 5K or a half marathon to predict my marathon?
The half, every time. Prediction error grows with the gap between the distance you raced and the one you are predicting, and a half marathon also tests fuelling and the ability to hold a pace for over an hour — the two things a 5K cannot tell you anything about.