How it works
T₂ = T₁ · (D₂/D₁)^1.06 (Riegel; + Cameron and VDOT)
Step 1 — enter your freshest hard result. A recent all-out race on an accurately measured course is the best input; if the course ran short or long, enter the distance it really was. Step 2 — read the consensus and the spread together. A 52:30 10K predicts a 1:56:00 half marathon, with the three models only 40 seconds apart (1:55:45–1:56:25). The same 10K predicts a 4:02:42 marathon, but now the models differ by 5:22 (4:00:37–4:05:58). Step 3 — if the goal race is a marathon, plan it more conservatively, because that is where the spread widens: a 24:10 5K gives 50:17 for the 10K and 1:51:07 for the half within a quarter-minute, but 3:52:37 for the marathon with 5:39 between the models. Step 4 — set the goal and the pace group. Start from the consensus, not the fast end. The 10K runner predicted at 1:56:00 sits between the 1:55 and 2:00 pace groups; 1:55 is a minute faster than the consensus and 45 seconds faster than even the most optimistic model, a fair stretch target only if the next ten weeks of training go well. A 1:58:00 half-marathon runner gets a 4:06:51 marathon consensus, and no model reaches 4:00 (the fastest says 4:03:46), so the 4:10 group is the sensible start. Step 5 — re-run after a tune-up race four to six weeks out and let the newer result replace the old one. For the physiology behind one of the three votes, look up your VDOT running score.
Sources
- Riegel endurance model Riegel, P. S. (1981). “Athletic Records and Human Endurance.” American Scientist 69(3), 285–290. Exponent 1.06 for adult runners.
- Cameron model David F. Cameron, race-time prediction model (1998), fitted to world-best times across 800 m–marathon.
- Daniels & Gilbert VDOT Jack Daniels & Jimmy Gilbert, “Oxygen Power: Performance Tables for Distance Runners” (1979); Jack Daniels, “Daniels’ Running Formula” (Human Kinetics).
FAQ
Which pace group should I start with?
The group at or just slower than your consensus prediction. Pace groups run even splits toward a fixed finish, usually in five- or ten-minute steps, so when your prediction falls between two of them, start with the slower one and move up only in the second half if the effort still feels controlled. Starting with a group faster than your prediction means running the first half beyond your current fitness, and most runners pay for that late in the race.
Should I race a tune-up before the goal race?
If the calendar allows, yes. A tune-up four to six weeks before the goal — a 10K or half before a marathon, a 5K before a 10K — gives the predictor fresher input that already reflects your training block. Re-run the numbers with it: if the new consensus is faster, you have evidence to nudge the goal; if it is not, keep the original plan and let race day decide.
What if my goal course is hillier or the forecast hotter?
The prediction assumes a course and weather similar to the race you entered, so a flat, cool 10K projected onto a hilly, warm half overstates what you will run. Adjust the goal before the start rather than at mile 8: use hill effort pacing for big climbs, plan a few percent slower for heat, and run the first half by effort.
My last race was on a short or long course — does it matter?
More than most runners expect. If your "10K" really measured 9.85 km, entering 9.85 km instead of 10 turns a 52:30 result from a 1:56:00 half-marathon prediction into 1:57:52, nearly two minutes slower. Certified courses are accurate; for informal or uncertified ones, use the measured distance or choose a different race as the input.
Does the prediction account for my age?
Not directly, and over an 8–12 week block it does not need to: all three models scale your own result, so the age and sex you raced at are already built in. Age matters when you compare performances across years or against other runners; for that, the age-grading calculator scores a result against age and sex standards.
Can I set a goal faster than the prediction?
Yes, within reason. The prediction captures your fitness on the day of the input race; ten weeks of consistent training can move it, but no formula can forecast by how much. A cautious approach is to aim slightly ahead of the consensus — the fast end of the range or the next pace group — and confirm it with a tune-up race, rather than choosing a goal several minutes beyond any model.
The predictions translate one result into equivalent times with three published models; they cannot see the training, taper, course or weather between now and race day. Treat the consensus as a starting goal to test in workouts and tune-up races, not as coaching, medical advice or a promise.