Age grading with the 2025 road tables, worked through

What age grading does

Age grading compares a race time with the best time expected for the runner's age and sex at that distance. The tables used here are Alan Jones' 2025 road age-grade tables, published on GitHub under a CC0 public-domain dedication, which the age grading calculator loads directly. For each sex, age (5 to 100) and road distance (road mile to 100K) there is one factor of 1 or less, and three numbers follow from it:

• Age standard = open standard / age factor. The open standard is the world record for the 5K, 10K, half and marathon; the other distances are fitted.
• Age grade = age standard / your time. In the tables' own words, 100% means you matched the open time for your age, and 50% means you took twice as long as a 100% runner.
• Age-graded time = your time x age factor: what the run would be worth for an open-class runner.

The documentation defines no level names such as "national class", so this guide does not use any.

What changed in the 2025 tables

From the repository's README and history document:

• The 2025 tables were approved on 10 January 2025 by the Masters Long Distance Running Council of USA Track and Field (they replace the 2020 tables, approved in May 2020).
• The single-age best performances behind them were again provided by Tom Bernhard.
• The standards are fitted at the 5K, 10K, half and marathon. For 2025 the author describes an improved interpolation for every other distance, on the logarithm of the distance: u = (log d - log d1) / (log d2 - log d1), and standard = S1 x (1 - u) + S2 x u.
• In July 2025 the men's standards were updated because the mile standards were too soft, giving many performances above 100%. The men's road-mile open standard in the version used here is 3:47 (227 seconds); a header in one sheet of the workbook still shows the older 3:52 (232 seconds).

The open standard alone is 5 seconds faster. Because the age grade is open standard / factor / time, that change by itself scales every men's road-mile grade by 227 / 232 = 0.978, about 2.2% of its value. The age factors were revised in the same update, so the exact change for a given age is not just that ratio; I have not compared the July file with the January one.

This site uses the version dated 27 July 2025.

One time, six ages

To see what the factors do, I graded one men's 5K time, 22:00, at six ages, and one women's 10K time, 55:00, the same way, all with the same code as the calculator.

Man, 5K in 22:00 (1,320 seconds). At 55: age standard 769 / 0.8425 = 912.8 seconds (15:13); age grade 912.8 / 1,320 = 69.15%; age-graded time 1,320 x 0.8425 = 1,112.1 seconds (18:32).

Woman, 10K in 55:00 (3,300 seconds). The women's open standard for the 10K is 1,726 seconds (28:46).

The first rows barely move: the factor is exactly 1 from age 19 to 29 for the men's 5K and from 21 to 27 for the women's 10K, so a 25-year-old is graded straight against the open standard. From the mid-thirties on, each decade takes about 7 points off the men's 5K factor (0.9824 at 35, 0.9125 at 45, 0.8425 at 55). The age where the factor leaves 1 differs by distance and sex: the women's marathon factor is 1 only at ages 26 and 27, the men's marathon from 22 to 35.

Equal age grades across distances

The calculator also shows what the same percentage means at every other road distance. I ran it for a 50-year-old man and a 60-year-old woman at 70%:

Two things show up. For the woman at 60, the half and marathon paces both come to 8:53 per mile, because her marathon factor (0.7965) is higher than her half factor (0.7718): the tables rate a 60-year-old woman's marathon relatively closer to the open standard. And the man's 70% marathon, 3:12:12, is eight minutes faster than the 3:20:07 that Riegel's formula predicts from his 20:52 5K.

That second point is the one to keep in mind. Equal age grades are not race predictions. They say what time would score the same percentage, not what time a runner can run. To plan a race from another race, use the race time predictor, and see why race predictors disagree for how cautious to be with the marathon.

Why ages 5 and 100 are edge cases

The factors are not read off the single-age bests one age at a time. The author's method document fits a five-part curve per distance (a parabola, a straight line, a flat section that contains the world record, another straight line and a final parabola), placed so that the standards are at or below every single-age best. At the two ends of the range, ages 5 and 100, it is the end parabolas that set the factor, and the resulting standards look like this:

Some of these are hard to read as "the best a person of this age can do". At age 5 the women's standard is faster than the men's at all 16 road distances (in the mile, 5:50 against 6:19), and in the mile the same happens again from 95 to 100. At every age from 10 to 94, at every distance, the men's standard is the faster one. At 100 the women's marathon factor is 0.0625, which sets the standard at almost 35 hours; between 95 and 100 that factor falls from 0.2158 to 0.0625. The RunScore text files distributed with the tables stop at 99, while the spreadsheet goes to 100.

None of this is an error in the calculator, which applies the published factors exactly. It is a reason to treat a grade for a young child or a centenarian as a rough comparison, and to prefer comparing a runner with their own previous grades.

Limits

These are road tables. Track races use different tables, which this site does not provide. A grade above 100% means a time faster than the standard for that age and distance: usually a wrong distance or time, occasionally a genuine record. Age on race day counts, in whole years. Courses differ in hills, weather and accuracy, so a grade on a fast course and one on a hilly course are not strictly comparable. To turn a grade into a pace, try the pace calculator.

Sources: Alan Jones, Age-Grade-Tables (2025 road tables, version 2025-07-27, CC0 1.0), GitHub; Riegel PS (1981), Athletic records and human endurance, American Scientist 69(3):285-290. Rules checked 7 October 2026. Found a problem? Tell us.

Questions