Does the 1.06 race-time rule hold? We tested it on today's world records
Our own calculation, 2026-10-30. The rule: Peter Riegel's endurance formula, T2 = T1 x (D2 / D1)^1.06, used by most race-time calculators (https://en.wikipedia.org/wiki/Peter_Riegel). The records: the current outdoor world records read from Wikipedia's "List of world records in athletics" on 2026-10-30 (https://en.wikipedia.org/wiki/List_of_world_records_in_athletics). Method: for each pair of distances, the exponent the records themselves imply, k = ln(T2/T1) / ln(D2/D1); and a least-squares fit of ln(time) on ln(distance) across 1500 m, mile, 3000 m, 5000 m, 10,000 m, half marathon and marathon. Second check by hand: men's 10,000 m to marathon, ln(7,170 s / 1,571 s) / ln(42,195 / 10,000) = 1.5182 / 1.4397 = 1.0545, as the script gives.
Key results:
- Men: the records fit an exponent of 1.0546 from 5000 m to the marathon (1.0626 including 1500 m and the mile). Women: 1.0451 (1.0571 with the short distances). Riegel's 1.06 is slightly too slow for the best runners, more so for women.
- Men's half marathon record (57:20) predicts the marathon at 1:59:32.16 with 1.06: the record is 1:59:30, 2.2 seconds off. From the 10,000 m record (26:11.00) 1.06 predicts 2:00:26.91: 56.9 s (0.79%) slower than the record. From 5000 m (12:35.36): 2:00:44.73, 74.7 s (1.04%) slow.
- Women: from the half marathon record (1:02:52) 1.06 predicts 2:11:04.36 against the 2:09:56 record, 68.4 s (0.88%) slow; from 10,000 m (28:54.14), 2:12:57.39, 181.4 s (2.33%) slow; from 5000 m (13:58.06), 2:13:57.92, 241.9 s (3.10%) slow.
- The exponent is not constant: from 1500 m to 5000 m the records imply 1.0792 (men) and 1.0787 (women), far above 1.06; from 10,000 m to the half marathon, 1.0498 and 1.0409.
Exponents implied by pairs of records
| From -> to | Men | Women |
|---|---|---|
| 1500 m -> 5000 m | 1.0792 | 1.0787 |
| 5000 m -> 10,000 m | 1.0564 | 1.0491 |
| 10,000 m -> half marathon | 1.0498 | 1.0409 |
| Half marathon -> marathon | 1.0596 | 1.0474 |
| 10,000 m -> marathon | 1.0545 | 1.0440 |
| 5000 m -> marathon | 1.0551 | 1.0457 |
The records used (2026-10-30)
| Distance | Men | Women |
|---|---|---|
| 1500 m | 3:26.00 | 3:48.68 |
| Mile | 3:43.13 | 4:06.91 |
| 3000 m | 7:17.55 | 8:06.11 |
| 5000 m | 12:35.36 | 13:58.06 |
| 10,000 m | 26:11.00 | 28:54.14 |
| Half marathon | 57:20 | 1:02:52 (mixed race) |
| Marathon | 1:59:30 | 2:09:56 (mixed race) |
What the exponent does to a recreational prediction (example, not a measurement)
A 50:00 10K predicts a marathon of 3:50:00.60 with 1.06, 3:56:43.75 with 1.08 and 4:03:38.69 with 1.10: each 0.02 adds about seven minutes. Which exponent fits an ordinary runner is not tested here: these are world records, run by athletes trained for each distance.
What this cannot tell you
- Records come from different athletes, years, courses and weather; a pair of records is not one runner's two races.
- The women's road records were set in mixed races, with male pacemakers, which Wikipedia marks; that may flatter the long distances.
- Nothing here says how recreational runners slow with distance; training volume for the longer race matters most and is not modelled.
For the reader to try
Inputs worth a slider: a recent race time and distance, the target distance, the exponent (1.03-1.12, default 1.06). Results: predicted time, pace per km, the change per 0.01 of exponent.