Race pace predictor
Enter a recent race. Get predicted times for every common distance plus your training paces. You get two numbers rather than one: the textbook formula every other calculator uses, and the steeper slowdown we measured across 22,422 Garmin predictions. At the marathon they can differ by half an hour.
Race predictions
Two curves. The textbook formula, and the steeper fade every athlete in our study actually carried.
- 1 mile4:02 /km6:295:42
- 5K4:19 /km21:3520:33
- 10KYour race4:30 /km45:0045:00
- Half marathon4:42 /km1:39:171:44:42+5:24
- Marathon4:54 /km3:27:013:49:17+22:17
- 50K4:57 /km4:07:494:37:49+29:60
Textbook is Riegel's 1.06 exponent, used by nearly every race calculator online. Observed is 1.131 — the median fade implied by Garmin's own predictions across 22,422 daily predictions. Every athlete in the cohort came out steeper than 1.06. That is what the watch assumes about you, not a proven truth: the evidence it is nearer reality is 11 real races, median error +0.6%, which is a case series and nothing more.
Training paces
Based on your current 5K race fitness.
- Easy / long runConversational, 75–80% of weekly volume5:32 /km
- Marathon paceSlightly slower than threshold4:58 /km
- Tempo / thresholdComfortably hard, 20–40 min4:37 /km
- 10K paceHard, sustainable for 30–50 min4:27 /km
- 5K / intervalRace effort, 3–5 min repeats4:19 /km
- VO2 max / repetition1500m–mile pace, short intervals4:03 /km
How this works
The Riegel formula
Peter Riegel published this equation in 1977 after analyzing hundreds of race times across distances. The 1.06 exponent accounts for the gradual slowdown as race distance grows.
T₂ = T₁ × (D₂ / D₁)^1.06It assumes you're similarly trained for both distances. A 5K specialist predicting a marathon will see a time that's optimistic — they lack the endurance base. A marathoner predicting a 5K will see a time that's pessimistic — they lack speed. The closer your race distances, the better the prediction.
Why the second column exists
We recorded Garmin's four race predictions every morning across a consenting cohort, 22,422 prediction-days in all, and solved for the exponent each athlete's own numbers implied. The median came out at 1.131, and every single athlete came out above 1.06. The half-marathon-to-marathon step was steepest at 1.164.
T₂ = T₁ × (D₂ / D₁)^1.131We keep Riegel as the first column because it is the honest default: it is what the rest of the internet gives you, and the second column is one cohort of self-selected Garmin owners, not a law of physiology. What we will not do is quietly swap in our own exponent and let you think it is the textbook.
Which is closer to what you will actually run is a genuinely open question. The only direct evidence we have is 11 races from the cohort, where the prediction standing the day before came within a median of 0.6%. That is a case series. It leans toward the steeper curve, and it settles nothing. Full method and limits.
Frequently asked questions
How accurate is a race pace predictor?
It depends almost entirely on how far you extrapolate. Between neighbouring distances (10K to half marathon) the Riegel formula holds up well. Across a big jump — 5K to marathon — it gets optimistic, and our own data says systematically so: across 22,422 daily Garmin predictions, every single athlete in our cohort implied a steeper slowdown than Riegel assumes, median 1.131 against his 1.06. That is why this calculator shows both curves instead of pretending there is one answer.
What is the Riegel formula?
Peter Riegel's 1977 formula: T2 = T1 × (D2 / D1)^1.06. The 1.06 exponent accounts for the fact that you slow down slightly per kilometer over longer races. It remains one of the most widely used and tested race predictors despite being almost 50 years old.
Can I use a treadmill or training run as input?
Only if it was a maximal effort over the stated distance. Race performance reflects pacing, fueling, and mental effort that regular training runs don't. Time trials and races are what the formula is designed for.
Why are my training paces different from what this predicts?
Training paces here are based on your current race fitness. If you train harder or easier than these paces feel, trust how you feel over the formula. These paces assume healthy recovery; in a high-stress week, easy pace should drift 15–20 seconds slower.
Does this work for trail or ultra running?
Not directly. Riegel assumes flat, consistent terrain. For trail races, add 10–30% to the predicted time depending on elevation gain and technicality. Ultras (50+ miles) break the formula entirely — fueling and pacing dominate.
Should I trust this or my Garmin race predictor?
They are answering slightly different questions. Garmin works from your actual training — VO2 max trend, recent paces, lactate threshold — while this works from one race result. The interesting part is that they disagree in a consistent direction: we measured the fade Garmin builds in and it is steeper than the textbook formula for every athlete we looked at. In 11 real races where we had a prediction from the day before, Garmin came within a median of 0.6%. Eleven races is a case series, not proof, but it is the only direct evidence we have and it does not favour the textbook.
Read more
A predicted time is one number. The writing below covers what sits underneath it — aerobic ceiling, zone distribution, and the research the formulas came from.