A 25:00 parkrun says you should be capable of 52:07 for 10K and 1:55:00 for a half marathon. It also says 3:59:47 for a marathon, and that last number is the one most runners get wrong. Race time predictors are reliable at short distances and quietly optimistic at long ones. Knowing where the line sits is worth about ten minutes on race day.
How does a race time predictor work?
A race time predictor takes one recent race result and scales it across other distances using the Riegel formula: T₂ = T₁ × (D₂ ÷ D₁)1.06. Pete Riegel, an American engineer and marathoner, published the idea in Runner's World in 1977 and set out the maths in American Scientist in 1981 after analysing endurance records across several sports.
That exponent of 1.06 carries the whole argument. If your speed held perfectly, the exponent would be 1.0 and doubling the distance would exactly double your time. It does not. Riegel found that in practice, doubling the distance multiplies your time by roughly 2.08, so pace drifts by about 13 to 15 seconds per kilometre each time the distance doubles.
The RunClub race predictor uses that exact formula. Feed it a 5K, 10K, half or a custom distance, and it returns finish times and target paces for the rest.
What a 5K time predicts across the distances
| Recent 5K | 10K | Half marathon | Marathon (Riegel) | Marathon (realistic) |
|---|---|---|---|---|
| 22:00 | 45:52 | 1:41:12 | 3:31:00 | 3:41:00 or slower |
| 25:00 | 52:07 | 1:55:00 | 3:59:47 | 4:09:47 or slower |
| 30:00 | 1:02:33 | 2:18:00 | 4:47:44 | 4:57:44 or slower |
Riegel figures calculated at exponent 1.06. The realistic marathon column adds the ten-minute correction explained below.
How accurate is a race time predictor?
Accurate up to the half marathon, unreliable at the marathon. That is the short answer, and it comes from real data rather than forum opinion. A 2016 study of 2,303 recreational endurance runners, published in BMC Sports Science, Medicine and Rehabilitation, tested the Riegel formula against actual race results.
It predicted 10K and half marathon times well. At the marathon it "dramatically underestimated" finishing time, producing predictions at least ten minutes too fast for half of the runners in the sample. The researchers' own models, which factored in weekly training mileage alongside race results, cut the mean squared error from 381 down to 228.
The lesson is not that the formula is broken. It is that a marathon punishes anyone whose weekly mileage does not back up their 5K speed. The formula only knows how fast you raced once. It has no idea whether you have run 30 miles a week for four months or 12 miles a week for three.
Race predictor or pace calculator: which one do you need?
They answer different questions, and mixing them up is why people end up with a number they cannot use. A race predictor answers "what could I run?" A pace calculator answers "what pace do I need to hold?" One is a forecast, the other is arithmetic.
| You want to know | Use |
|---|---|
| My realistic 10K time from a recent parkrun | Race predictor |
| The per-km pace for a sub-2:00 half | Pace calculator |
| Whether a goal marathon time is plausible | Race predictor |
| My pace in min/mile when I train in min/km | Pace calculator |
| Kilometre splits to write on my hand | Pace calculator |
In practice you use both, in order. Predict the time first, then take that time to the pace calculator to convert it into a per-kilometre and per-mile pace you can actually run to. Our post on what pace to run a 10K covers the second half of that job in more detail.
What to feed a race predictor so the answer is worth having
Rubbish in, rubbish out. The formula is only as honest as the race you give it, and three things decide that.
Recency. Use a result from the last six to eight weeks. A personal best from 2023 predicts the fitness of a person who no longer exists.
Effort. It has to have been a genuine race, run flat out. A social club run at 90 percent effort will underrate you; a downhill course with a tailwind will flatter you.
Distance gap. Predictions get shakier the further you extrapolate. A 10K predicting a half is solid. A 5K predicting a marathon is a guess with a decimal point.
The easiest honest input in Britain is a parkrun, free, timed and on the same 5K course every Saturday morning, so you can repeat it and watch the trend rather than trusting one result.
How to correct a marathon prediction before it ruins your race
Take whatever the predictor gives you for 42.2km and add at least ten minutes. That single adjustment brings you in line with what the 2016 study actually observed, and it costs you nothing if you turn out to be fitter than expected.
Then sanity-check it against your training. If the predicted marathon pace is not a pace you have held for at least 90 minutes in a long run, it is a fantasy figure. Runners who blow up at 30km almost always set their goal from a sharp 5K rather than from their longest run. Build the test properly: use the route creator to map a flat 20km loop, then hold the predicted pace for the whole thing and see what your legs say.
Autumn is when this bites in the UK. If you are aiming at a September or October race, you have roughly eight weeks of long runs left to prove or disprove the number. Our Great North Run first-timer guide walks through the same pacing logic over 13.1 miles, where the formula is on much firmer ground.
Frequently asked questions
How do I predict my marathon time from a half marathon?
Multiply your half marathon time by 2.08 for the Riegel estimate, then add ten minutes. A 1:50 half gives 3:49 on the formula and around 3:59 realistically. High-mileage runners land nearer the raw figure; runners training under 30 miles a week rarely do.
Is a parkrun time good enough to predict a race?
Yes, if you raced it. A hard parkrun is a legitimate 5K time trial and the most repeatable data point most UK runners have. Just check the course profile, because a hilly or trail parkrun will read slower than your road fitness deserves.
Why does the predictor say I should run faster than I can?
Almost always training volume. The formula assumes you are equally prepared for every distance. If your weekly mileage suits a 5K, the marathon prediction will overshoot, because endurance, fuelling and leg durability are not captured by one race result.
How often should I re-run the prediction?
Every four to six weeks during a training block, using your most recent hard effort. More often than that and you are measuring noise, weather and how well you slept rather than fitness.
Test the number with people, not just a calculator
The fastest way to find out whether a predicted pace is real is to run it alongside someone who already holds it. Most clubs have a Tuesday session organised around exactly that, and a paced group run will tell you more in 45 minutes than a fortnight of solo guessing. Find a running club near you and take your predicted pace out for a test drive.
