Race Time Predictor
Predict race times across distances using the Riegel formula. Enter your known race, get predictions for 1 mile, 5K, 10K, half, marathon, 50K, 100K. The exponent is adjustable.
Race Time Predictor
| Distance | Predicted Time | Pace |
|---|---|---|
| Enter your known race distance and time above. | ||
How to use the Race Time Predictor
Enter one race you have run
Type its distance in kilometres (up to 200) and your finishing time. Blank hours, minutes or seconds count as zero.
Read the table
Each row is your time scaled by Riegel's formula to another distance, with the pace per km. Your own race is highlighted. The 50K and 100K rows are marked, because they extend the formula beyond the marathon.
Try another exponent if you want
The default is 1.06, the value the Runner's World online calculator uses (as Vickers and Vertosick report). The box accepts 1.00 to 1.20 and refuses anything else, rather than quietly swapping in a different number. The line under the table prints the exponent actually used, and what doubling the distance does to the time at that exponent.
Read the result as the formula's answer
The table is what the formula computes from your one race, nothing more. The sections below set out how close that comes for world records and for recreational marathon runners.
Riegel formula — predicting one race from another
The calculator scales a known time by a power of the distance ratio: T₂ = T₁ × (D₂ / D₁)^k. P. S. Riegel set out this power law in "Athletic records and human endurance" (American Scientist, 1981). We could not open the paper itself. As Vickers and Vertosick (2016) report it, "Riegel cited a k of 1.08 for elite runners and 1.05 to 1.06 for male recreational runners aged 40 to 70", while the Runner's World online calculator uses 1.06. So 1.06 is one of the values Riegel reported, used on its own for every runner and distance. It is this tool's default.
What 1.06 does to the numbers
With an exponent of exactly 1, pace would not change with distance. At 1.06, doubling the distance multiplies the time by 2^1.06 = 2.085, which covers both 5K to 10K and half marathon to marathon (each an exact doubling). Going from 10 km to the marathon multiplies it by 4.2195^1.06 = 4.600.
Riegel fitted his curve to world records. For everyone else it is a starting estimate, not a promise.
Against the world records
Run on the World Athletics record list as it stood on 15 October 2026, the formula holds up well. From Cheptegei's 26:11.00 for 10,000 m, 1.06 gives a 2:00:27 marathon, 57 seconds slower than Sawe's 1:59:30. That is the closest pair. The women's 10,000 m record (Chebet, 28:54.14) scaled the same way gives 2:12:57, 2.0% faster than the women-only marathon record and 2.3% slower than the mixed-race one. Each pair is two different athletes on two different days, so this tests the curve, not any one runner.
Past the marathon the curve runs out. Scaled from the men's marathon record, 1.06 gives a 100 km time of 4:58:15, 18.4% faster than Sorokin's record of 6:05:35. That is why the 50K and 100K rows are marked.
Where it misses: recreational marathons
Vickers and Vertosick surveyed recreational runners through Slate.com in 2014 and published the data with their paper. Using 1.07, they found Riegel's formula "was well-calibrated for races up to a half-marathon, but dramatically underestimated marathon time, giving times at least 10 min too fast for half of runners." We re-ran their published data at this tool's 1.06:
| Known race → predicted race | Runners | Median miss at 1.06 |
|---|---|---|
| 5 km → 10 km | 500 | 1.3% too fast |
| 10 km → half marathon | 529 | 1.1% too fast |
| 5 km → marathon | 430 | 6.9% too fast |
| 10 km → marathon | 289 | 5.5% too fast |
| Half marathon → marathon | 717 | 5.3% too fast |
For 10 km to the marathon, the median miss was about 12 minutes. About four in five runners were predicted too fast, about half by more than 10 minutes, and only about a quarter landed within 3% either way. The rows overlap, because one runner can appear in several pairs. These are self-reported times from a self-selected, mostly US readership in 2014. We found no comparable open data for races in Southeast Asia.
The exponent box
Vickers and Vertosick chose 1.07 as an average. In their data, the exponent implied by each runner's own two races had a median of about 1.08 for 5 km to 10 km, 1.07 for 10 km to the half and 1.14 for half to marathon. Changing the box shows how far the answer moves. It cannot tell which value describes any one runner.
What 1.06 was fitted to, and how far recreational marathons stray from it
T₂ = T₁ × (D₂/D₁)^k: Riegel's power law (American Scientist, 1981). Calculators use k = 1.06.
Riegel is reported to have cited about 1.08 for elite runners and 1.05 to 1.06 for male recreational runners aged 40 to 70 (Vickers and Vertosick, 2016).
At 1.06, doubling the distance multiplies the time by 2.085, not 2.
10 km → marathon multiplies the time by 4.600 at 1.06.
From the men's 10,000 m record, 1.06 lands 57 seconds off the men's marathon record. Record pairs from 10 km to the marathon miss by up to about 2.5% either way.
From the men's marathon record, 1.06 puts the 100 km record 18.4% too fast.
For recreational runners, 5 km → 10 km and 10 km → half predictions had median misses near 1%.
10 km → marathon: median 5.5% too fast, and about four in five runners predicted too fast.
Only about a quarter of recreational marathon predictions landed within 3% either way.
The formula has no input for hills, terrain or weather: it scales one time by one distance ratio.
Frequently Asked Questions
-
It depends on who and how far. Between world-record pairs from 10 km to the marathon it lands within about 2.5% either way. For recreational runners in the Vickers and Vertosick survey data, predictions up to the half marathon had median misses near 1%. At the marathon, the median prediction was about 5.5% too fast, and only about a quarter of runners finished within 3% of it.
-
The formula uses one exponent at every distance. In the recreational survey data, the exponent implied by runners' own half-marathon and marathon times had a median of about 1.14, well above 1.06. Up to the half marathon it was closer to 1.07 or 1.08. The exponent jumps at the marathon, and a single fixed value cannot follow that jump.
-
In the survey data at 1.06, the median marathon miss was 6.9% too fast from a 5 km, 5.5% from a 10 km and 5.3% from a half marathon. Each figure comes from a different, overlapping group of runners, so read them as a rough ordering, not a ranking for any one person.
-
The formula has no input for climb, terrain or weather, so it cannot tell a trail race from a flat road. For ultras, scaling the men's marathon record to 100 km at 1.06 comes out 18.4% faster than the actual 100 km record. That is why the 50K and 100K rows carry a "beyond the marathon" mark.
-
Yes. Dave Cameron's model (posted to the t-and-f mailing list in 2001) fits speed rather than time against distance, so its scaling factor depends on the two distances, not on how fast you ran. On the same survey data it cut the 10 km to marathon median miss from 5.5% to 3.8% too fast. It does worse on records, though: it puts the men's 10,000 m to marathon pair 2.7% too slow, against 0.8% too slow for Riegel. Vickers and Vertosick fitted their own regression models in the 2016 paper, which also use weekly mileage. Those models lowered the marathon mean squared error from 381 for Riegel to 228 with one prior race and 208 with two. A 2025 study of Valencia finishers (Oficial-Casado et al.) compared a half-marathon-based regression with Jack Daniels' VDOT tables. Those errors are measured differently, so they cannot be set beside the figures above.
-
Not as it stands. The 1.06 default and every check on this page come from running records and runners' results.
-
No. The prediction is calculated in your browser, and the distance and time you type are not sent anywhere. The page itself still loads the site's usual ads and analytics, like any other page.
Method & sources
How it computes
Riegel's power law, T2 = T1 x (D2 / D1)^k, default k = 1.06, adjustable 1.00-1.20 (anything else is refused). Riegel is reported to have cited about 1.08 for elite runners and 1.05-1.06 for male recreational runners aged 40-70 (Vickers & Vertosick 2016). Times rounded to whole seconds before splitting. RaceTimeKernel, tested.
What this tool implements
- Checked against the World Athletics record list (15 Oct 2026): from the men's 10,000 m record, k = 1.06 gives a marathon 57 s slower than the record; women's 10,000 m to marathon is 2.0% fast (women-only record) or 2.3% slow (mixed race); men's marathon to 100 km is 18.4% fast. Our arithmetic.
- Recreational runners: Vickers & Vertosick (2016) found Riegel well calibrated up to the half marathon but at least 10 min too fast at the marathon for half of runners (k = 1.07). Our re-run of their published data at 1.06: 10 km to marathon median 5.5% too fast, about a quarter within 3%.
- Corrected 2026-10-15: removed "within 3-5%" accuracy, "1.05 for elites", heat-penalty and temperature-exponent figures, and "used by Strava, Garmin, COROS" — none had a source.
Sources
- Riegel PS. Athletic records and human endurance. Am Sci 1981;69(3):285-90. PMID 7235349 (cited; not opened)
- Vickers AJ, Vertosick EA. An empirical study of race times in recreational endurance runners. BMC Sports Sci Med Rehabil 2016;8:26. https://doi.org/10.1186/s13102-016-0052-y
- Cameron D. Race-time prediction formula, t-and-f mailing list, 20 Jun 2001. https://www.mail-archive.com/[email protected]/msg11312.html
- World Athletics. World records by category (read 15 Oct 2026). https://worldathletics.org/records/by-category/world-records
Pick up where you left off
Stored only in this browser — never sent to our servers.