Running Race Predictor

It depends on your distance and training. Riegel (1977) is the textbook formula and works well in the 5K to half-marathon band for well-trained runners. Cameron (1998) tracks elite times slightly better above the half. Daniels' VDOT is most reliable when your reference race is 15 to 50 minutes long. The calculator shows all three side by side so you can see how much agreement there is. If you've entered two or more races it also fits your personal fatigue exponent, which is usually the closest match to your real numbers.

Riegel's classic formula assumes everyone slows down at the same rate as distance grows — a fixed exponent of 1.06. In reality the exponent varies. Speed-leaning runners (track and 5K specialists with shallow aerobic development) often run an exponent of 1.10 or higher. Diesel-engine marathoners can sit at 1.04 or even below. With two or more race results the calculator fits your own exponent from your data, so the predictions reflect your strengths rather than the population average.

VDOT (sometimes written VO2-dot) is the pseudo-VO2max that Jack Daniels and Jimmy Gilbert proposed in Oxygen Power (1979). It combines two things into one number: how much oxygen your body can take in (aerobic capacity) and how efficiently you convert that oxygen into running speed (running economy). Two runners with the same lab-measured VO2max can have very different VDOTs because of economy. The calculator derives VDOT from your strongest race, then uses it to set your training paces.

Running uphill costs extra oxygen per metre climbed — about 1.31 mL per kg per metre, from Minetti's published treadmill data. The calculator converts that extra oxygen demand into extra seconds at your race-pace metabolic rate, then adds it to the predicted time. Net downhill courses give back some time, but only about half what equivalent climbing would cost, because braking forces and gait limits cap how fast you can descend. This is the same physiology that drives Strava's Grade Adjusted Pace.

Above about 10°C, pace gets progressively slower as the temperature rises. The calculator uses the curve from El Helou et al. (2012), who fit 1.8 million finish times from six major marathons. Slowdown is roughly +1% at 15°C, +3% at 20°C, +8% at 25°C, and +14% at 30°C. Slower runners get hit harder than faster ones because they're exposed to the heat for longer — these numbers are the average across recreational finishers. Acclimation also matters: regular heat training can roughly halve the penalty.

These are the five training intensities from Daniels' Running Formula. Easy/Long runs sit at 65 to 79% of VDOT — the conversational pace that builds your aerobic base. Marathon pace is about 80% VDOT, used for goal-pace work. Threshold (also called tempo or T pace) is about 88%, where you train your ability to clear lactate. Interval (I pace) is around 98%, used for VO2max repeats of 3 to 5 minutes. Repetition (R pace) is about 105% VDOT, short and fast for speed and economy.

One race is enough to get Riegel, Cameron and VDOT predictions. Two or more lets the calculator fit your personal fatigue exponent, which usually fits your real history better than the textbook 1.06. The fit works best when your races span a wide distance range — a 5K and a half-marathon tells the model far more than two 10K times. Three or more inputs also gives you an R² so you can see how cleanly the points line up.

A linear extrapolation (just multiplying the per-km pace) would assume you can hold 10K pace for over four times the distance, which nobody can. Riegel's exponent of 1.06 says doubling the distance costs about 4% more than double the time, and a marathon is roughly 4.2× a 10K — so the predicted marathon time is closer to 4.65 × 10K time, not 4.22 ×. The Cameron and personal-fit numbers give slightly different answers, but they all sit well above the linear figure.

Within the last 6 to 12 weeks for the best signal. Fitness shifts with training blocks, illness, taper status and altitude. A six-month-old PR is more aspirational than predictive. If you've had a clear fitness change since a race, drop it from the inputs or replace it with a recent time trial.

Net elevation gain, yes — enter the total climb on your target course. Surface (road vs. trail) and technicality aren't modelled directly. For trail races the rule of thumb is to take the road-equivalent prediction and add 10 to 30% depending on terrain. The elevation adjustment captures the energetic cost of climbing but not the time you'll lose to picking lines or hopping rocks.

No. Everything runs in your browser. Your race inputs and preferences are saved locally.