Running Pace Calculator: Pace, Splits and Race Predictions

A pace calculator does two things. It converts a distance and a finishing time into a rate, expressed as time per unit of distance, and it projects that rate onto other distances using an empirical endurance formula. The first part is division. The second part is a model, and models have limits.

Runners describe effort in minutes per kilometer or minutes per mile rather than in kilometers per hour, because pace maps directly onto what happens on a watch. If you know you are running 5:00 per kilometer, you know each kilometer marker should tick past five minutes apart. Speed in km/h requires a conversion in your head at exactly the moment you are least able to do one.

This page is about the arithmetic of pacing. It is general information, not coaching and not medical advice. A predicted time is a number produced by a formula from one prior result; it is not a training plan and carries no implication about what any individual can or should attempt.

#The formula

Formula: Pace = time / distance. Speed = distance / time. In consistent units: pace in seconds per km = total seconds / distance in km, and speed in km/h = distance in km / (total seconds / 3600).

Convert the finishing time to total seconds first, since mixed hours, minutes and seconds do not divide cleanly. A time of 50 minutes is 3,000 seconds. The same hours-minutes-seconds arithmetic appears in the time card calculator.

To convert the result back into pace notation, divide the seconds-per-unit figure by 60 for whole minutes and take the remainder as seconds. 300 seconds per km is 300 / 60 = 5 minutes exactly, so 5:00 per km.

Pace and speed are reciprocals of one another, scaled by the units. Per-km pace in seconds multiplied by speed in km/h always equals 3,600.

#A worked example, done by hand

Take a 10 km race finished in 50:00.

1. Total seconds: 50 x 60 = 3,000.
2. Pace per km: 3,000 / 10 = 300 seconds, which is 5:00 per km.
3. Speed: 10 / (3,000 / 3,600) = 10 / 0.8333 = 12.0 km/h.
4. Distance in miles: 10 / 1.609344 = 6.2137 miles.
5. Pace per mile: 3,000 / 6.2137 = 482.8 seconds, which is 8:03 per mile.
6. Speed in mph: 6.2137 / 0.8333 = 7.46 mph.
7. Even-split halfway: 3,000 / 2 = 1,500 seconds, or 25:00 at 5 km.

Every other figure on the page is a transformation of those.

#Pace versus speed

Speed is distance per unit time. Pace is time per unit distance. They contain identical information and are inverses, but they behave differently in the hand.

Pace subtracts and adds cleanly against a course. If you are 20 seconds behind at 5 km, you know you need to find 4 seconds per km over the remaining five. Speed does not decompose like that.

Pace also has a useful asymmetry: at the fast end, small pace changes represent large fitness changes, while at the slow end they do not. Moving from 4:00 to 3:50 per km is a bigger step than moving from 7:00 to 6:50, and pace notation makes the relative size of the change visible in a way that a treadmill speed dial does not.

The one place speed wins is equipment. Treadmills are calibrated in km/h or mph, which is why the calculator reports both.

#Converting between per-km and per-mile pace

One mile is 1.609344 km, so per-mile pace is per-km pace multiplied by 1.609344. The rough mental version is to add about 60 percent.

Note that a 30-second gap in per-km pace becomes roughly a 48-second gap per mile. Mixing units within a single session is the most common source of confusion in pacing arithmetic, particularly on a treadmill set to one unit while a watch reports the other.

#Riegel's endurance formula

Formula: T2 = T1 x (D2 / D1)^1.06, where T1 is a known time over distance D1, and T2 is the predicted time over distance D2. Both distances must be in the same unit.

Peter Riegel published this in the late 1970s after fitting a power curve to race results across a range of distances. The structure says that time does not scale linearly with distance: doubling the distance multiplies the time by 2^1.06, which is about 2.085, not 2. The extra 4 percent is the cost of endurance.

The exponent is an empirical fit, not a physical constant. Riegel found values close to 1.06 across running events, and the figure has stuck because it is simple and works reasonably over moderate extrapolations.

Applying it to the 10 km in 50:00:

The marathon prediction is the calculation stretched furthest, and it deserves the least confidence.

#Where the formula stops working

Riegel's fit was drawn from race results, which means it describes runners who were prepared for the distances they ran. Extrapolating past roughly a doubling of distance assumes an equivalent level of preparation for a race you may not have trained for, and the model has no way to know that.

For a first marathon, this matters more than anywhere else. Beyond about 30 km, fuel depletion, muscular fatigue and accumulated mechanical damage change the problem in a way that a smooth power curve cannot represent. A 10 km time is a measurement of a race lasting under an hour, and it says relatively little about what happens in hour four.

The sensitivity is easy to see by varying the exponent on the same 10 km input:

Thirteen minutes separate the first two rows and thirty-two the first and last, from a parameter change most people would call small. Many runners find that a higher exponent describes their own long-distance results better, particularly when the base result comes from a short race. Comparing predictions from a 5 km, a 10 km and a half marathon is more informative than trusting any single one, and if they disagree widely the shorter inputs are usually the optimistic ones.

#Splits: even, negative and positive

A split is the time for a section of the race. The calculator's halfway figure assumes even effort: half the total time at the halfway point.

Even splits mean each half takes the same time. Simple, and a reasonable default for a well-judged race.

Negative splits mean the second half is faster than the first. This requires starting slower than average pace and holding something back.

Positive splits mean the second half is slower. This is what happens when the opening pace was too fast, and the loss at the end usually exceeds the time gained at the start.

For a 4:00:00 marathon, even pace is 14,400 / 42.195 = 341.3 seconds per km, which is 5:41 per km or 9:09 per mile:

For the 10 km example, even splits are 5:00 per km throughout, with 25:00 at halfway. A two-minute negative split over that distance would mean 26:00 for the first 5 km and 24:00 for the second, which on a 50-minute race is a large swing.

Note that the halfway point by distance is not the halfway point by effort on any course with hills, and the split table assumes it is.

#What breaks a prediction

Course profile. Elevation gain costs time that flat-course arithmetic does not model, and the descent rarely returns it in full. A rolling course and a flat one can differ by minutes at the same effort.

Surface. Trail, sand, grass and wet ground all cost pace relative to asphalt.

Heat and humidity. Thermoregulation competes with locomotion for blood flow. Warm, humid conditions degrade sustainable pace, and the effect grows with race duration.

Altitude. Reduced oxygen availability lowers sustainable pace at endurance intensities, with the size depending on elevation and acclimatization.

Wind. Headwind costs more than an equal tailwind gives back, because drag rises with the square of relative speed.

Crowding and course furniture. Congested starts, tight turns and aid stations all add seconds that no formula includes.

Riegel's fit describes flat, temperate, well-organized races. Every departure from that moves real times one direction only.

#How to read the result

The pace figures are exact arithmetic and can be trusted completely. The projections are estimates from a curve fitted to other people's races, and should be read as a range rather than a target.

A sensible reading: treat a projection from a similar distance as a reasonable reference point, and treat a projection more than double the input distance as an optimistic bound. If a prediction and your own long-run experience disagree, the experience is the better evidence.

A prediction is not a training prescription. It says what the formula implies from one result, and it knows nothing about training history, injury, recovery, sleep or anything else that determines what happens on a given day. Decisions about training load and racing belong with a coach, and any question about your health belongs with a clinician. The BMI calculator and calorie calculator are likewise arithmetic tools rather than health guidance. The disclaimer sets out the limits of everything on the site, and the full tool list is here.

#What this model leaves out

• Training status. The formula assumes comparable preparation for every distance.
• Fueling. Beyond roughly two hours, energy availability becomes a limiter the model does not represent.
• Terrain and weather. Every projection is a flat-course, temperate-conditions figure.
• Pacing discipline. The formula predicts a finishing time, not the ability to execute a pace.
• Recovery state. A time set fresh and a time set tired are treated identically.
• Individual profile. Some runners are relatively stronger at short distances, others at long ones, and one exponent cannot describe both.
• Race measurement. Course tangents mean recorded distance is often slightly longer than the official distance, which shifts real pace against watch pace.

#Common mistakes

Mixing units. A per-mile pace applied to kilometer markers runs about 60 percent slow. Fix the unit before anything else.

Converting pace as a decimal. 5:30 per km is 5.5 minutes, not 5.30. Convert to seconds, multiply, convert back.

Trusting a marathon projection from a 5 km time. That is more than an eightfold extrapolation, and it is where the formula is weakest.

Pacing off average speed on a hilly course. Even effort and even pace are different things once there is elevation.

Forgetting the tangent problem. Watches routinely record a marathon as longer than 42.195 km, so watch pace reads slower than official pace even in a perfectly run race.

Treating the halfway split as a plan. It is arithmetic showing what even effort looks like, not an instruction.

Comparing predictions across different formulas. Different exponents and different fitting data give different answers, and none of them is a measurement.