Pace calculator
The race times below are the same pace extended to longer distances, which is not how running works: everyone slows as the distance rises. Riegel put a number on it. Time scales with distance to the power 1.06, so each doubling costs about 4.25% on the pace, and a 50-minute 10K projects to 1:50:19 for a half and 3:50:01 for a marathon rather than the 1:45:29 and 3:30:59 below. Even Riegel runs optimistic at marathon distance. Treat these rows as a floor, not a forecast.
The pace, time and distance arithmetic is exact. The projected race times are not: they hold one pace across every distance, which nobody does, so read them as the best case rather than a target. Anyone stepping up to a first half or first marathon should build the distance gradually and speak to a doctor before starting hard training after a long break.
Pace is time divided by distance. Ten kilometres in 50 minutes is a pace of 5:00 per kilometre, or 8:03 per mile, at 12 km/h. Any two of pace, time and distance give the third.
How to calculate running pace
Half of this panel is exact and half of it is a guess, and the boundary runs between the top rows and the bottom ones. Pace, time, distance and speed are one identity with three unknowns; given any two, the third follows with no assumptions at all. The mile is 1.609344 km by definition, so a mile pace is always the kilometre pace times 1.609. That catches people out because it is not a conversion you can do in your head: 5:00/km is 8:03/mile, not 8:00. Speed in km/h is the view treadmills use, and 5:00/km is exactly 12 km/h, one of the few places where the arithmetic comes out clean.
The projected race times below are a different kind of number. They extend one pace across every distance, and nobody holds one pace across every distance. Peter Riegel proposed the standard correction in 1981: time rises with distance to the power 1.06, so each doubling costs about 4.25% on the pace. Applied to a 50-minute 10K that turns a 3:30:59 marathon into 3:50:01, which is a large difference to discover on the day.
Even the correction is optimistic at the long end. Vickers and Vertosick surveyed 2,303 recreational runners and tested Riegel against their actual results. It was well calibrated up to the half marathon and dramatically underestimated marathon time, giving predictions at least ten minutes too fast for half the sample. Their own models, built from one or two prior race times plus training volume, cut the mean squared error from 381 to 228. The practical reading is that a projection from a shorter race is a floor: the pace you could hold if nothing about the longer distance were harder, which is never true, and least true over 42 kilometres.
What people use it for
- Planning a target pace for a race
- Converting a treadmill speed to a pace
- Working out how far you can go in a set time
- Comparing training paces across units
- Sanity-checking a race projection before committing to it
Questions
Divide total time by distance. Fifty minutes over 10 km is 5:00 per kilometre.
8:03 per mile. Multiply the kilometre pace by 1.609.
Sixty divided by the pace in minutes. A 5:00/km pace is 12 km/h.
They assume you hold the same pace at every distance, which nobody does. Riegel gets closer: time rises with distance to the power 1.06, or about 4.25% on the pace for each doubling.
T₂ = T₁ × (D₂ ÷ D₁)^1.06. A 50-minute 10K projects to 1:50:19 for a half marathon and 3:50:01 for a marathon, against the 1:45:29 and 3:30:59 the rows above give by holding the pace flat.
Not really, and there is a number on that too. Vickers and Vertosick surveyed 2,303 recreational runners and found Riegel well calibrated up to a half marathon but dramatically optimistic for the marathon, giving times at least ten minutes too fast for half of the runners in their sample. Both the flat projection and the Riegel correction should be read as best cases.
Pace, time, distance and speed. They are one identity and involve no assumptions. Everything from the projected 5K downwards is a model of how runners slow over distance, and models can be wrong.
Most easy running should be comfortably conversational: considerably slower than race pace, and slower than most people actually run it.