Running pace calculators look simple from the outside. You type in a recent race result and they spit out a predicted marathon time. But the math behind that output has a history grounded in exercise physiology, empirical observation, and decades of testing against real race results.
Understanding how these models work helps you use them more accurately. It also helps you understand why a prediction might be off, and what you can do to get better inputs.
The Foundation: Riegel's Formula
The most widely used race prediction formula comes from Peter Riegel, who published it in a 1977 paper in American Scientist. The formula is:
T2 = T1 * (D2 / D1)^1.06
Where T1 is your known time at distance D1, and T2 is the predicted time at distance D2. The exponent 1.06 is the critical term. It captures the empirical observation that running performance does not scale linearly with distance. The longer the race, the slower your pace becomes relative to your speed at shorter distances.
The 1.06 exponent was derived from observing how race times distributed across distances for a large population of runners. It is not a theoretically derived constant from exercise physiology. It is a fitted value from observed data, which means it works well on average but less well at the extremes.
Why the Formula Works for Moderate Distances
Riegel's formula is most accurate for predictions between adjacent distances: 5K to 10K, 10K to half marathon, half marathon to marathon. The error rates are small enough to be practically useful.
For example: a 22:00 5K runner (4:24 per kilometer) can expect a 10K somewhere around 45:44 using Riegel's formula. Actual times for runners with that 5K baseline cluster within a minute or two of that prediction, assuming the runner is similarly prepared for the 10K distance.
Where the formula degrades is at very long distances and for runners who specialize heavily at one distance. A runner who does nothing but 5Ks will have a 5K that overpredicts their marathon. Their anaerobic and short-distance fitness is high, but their aerobic base and fat metabolism at marathon pace have not been developed. The formula assumes a generalist athlete.
Photo by ThisIsEngineering on Pexels
VO2 Max Estimation as an Alternative Approach
Some prediction tools take a different approach. Instead of fitting directly to race times, they estimate VO2 max, your maximal rate of oxygen consumption per kilogram of body weight, and then predict performance from that physiological measure.
The relationship between VO2 max and running performance is well established. Higher VO2 max generally correlates with faster times, especially at middle distances. The challenge is that VO2 max is hard to measure directly. Lab testing requires a metabolic cart and a trained technician. Field estimates are approximations.
One commonly used field estimation method: if you run a timed mile, your VO2 max is approximately:
VO2max = 3.5 * (483 / time_in_minutes + 3.5)
This estimate has real uncertainty bounds, but it gives a usable ballpark. From an estimated VO2 max, you can reverse-engineer predicted race times using tables developed by physiologists like Jack Daniels, whose VDOT system has been used in structured training programs for decades.
VDOT stands for velocity at which you achieve VO2 max adjusted for running economy. It is not exactly the same as VO2 max but correlates with it closely. Daniels' tables map VDOT values to training paces and race times across standard distances, which is why you sometimes see calculators that output easy pace, tempo pace, and interval pace from a single race input. They are computing VDOT from your race time and then reading training paces off the table.
The Role of Running Economy
VO2 max tells you how much oxygen your body can use. Running economy tells you how efficiently you use it. Two runners with identical VO2 max scores can have meaningfully different race times if one is a more economical runner.
Running economy is influenced by factors like stride mechanics, ground contact time, vertical oscillation, and body composition. It is trainable, and it often improves with volume even when VO2 max plateaus. This is why experienced runners sometimes run times that prediction models underestimate: their economy has improved without a corresponding jump in raw aerobic capacity.
Most consumer prediction tools do not account for running economy because it requires sensor data that wrist-based devices cannot reliably measure. Advanced training platforms from Garmin and Polar attempt to estimate economy from accelerometer and optical sensor data, but these estimates have significant uncertainty.
How Temperature and Elevation Are Factored In
Basic pace calculators take time and distance as inputs and produce a prediction assuming standard conditions. More sophisticated tools allow you to adjust for temperature and elevation.
Temperature adjustment models typically apply a performance penalty of 1 to 2 percent per 10 degrees Fahrenheit above 60 degrees at race start. This is based on research showing that cardiovascular drift, the gradual increase in heart rate at a constant pace during heat exposure, produces measurable slowing across a race.
Elevation adjustment is more complex. Net elevation gain over a course produces a pace penalty. Net descent produces a pace benefit, though the benefit is smaller than the penalty at equivalent grades because the braking forces in downhill running are costly. A rough adjustment for net gain is 30 to 45 seconds per mile per percent of average grade.
These adjustments are estimates, not exact science. Races with significant elevation change or unusual weather conditions will always have more prediction uncertainty than flat, mild-weather events.
What Calculators Cannot Account For
Race preparation. Riegel's formula assumes you are equally prepared at both distances. If you have run ten 5Ks this year but have only one 10-mile long run under your belt, your predicted marathon time from a 10K is going to be optimistic.
Tapering and freshness. A prediction derived from a training run does not account for the fitness and freshness of a proper taper. Runners often run times at major races that outperform their best training predictions because they arrive rested.
Mental execution. Race day pacing decisions, particularly going out too fast, account for a large share of positive splits and missed goals. A calculator gives you the correct starting pace. Whether you actually run it is a different matter.
Individual variability. The exponent in Riegel's formula fits the population well, but individual runners can deviate by several percentage points. A runner with unusually high slow-twitch fiber density might be a better marathon predictor at a higher exponent. A runner with more fast-twitch development might be worse.
Using a Pace Calculator Well
The best way to use a pace calculator is to update your predictions regularly from recent race results and to treat predictions as probability ranges rather than single point estimates. A 5K from January means something different in April if you have done twelve weeks of structured training.
EvvyTools has a free running pace calculator that uses the Riegel formula to predict finish times and output training paces. The companion article Running Pace and Race Time: How the Math Works explains the model in more detail if you want to understand the math behind the predictions.
For the academic background, Wikipedia's article on VO2 max covers the physiological foundations. Jack Daniels' VDOT methodology is detailed in his book and widely referenced in structured training plans for all distances.
Photo by jonas mohamadi on Pexels
The Bottom Line
Pace prediction is pattern matching against a large population of runners using a relatively simple power law. It works well for most people at moderate distances under typical conditions. Its errors are mostly predictable: overestimation when you specialize at short distances, underestimation when you have strong aerobic economy, and significant drift when race conditions deviate from the training conditions that produced the input.
Use the prediction to set your opening pace. Trust your training. Adjust in real time when conditions warrant.
Top comments (0)