How to Predict Your Marathon Time

Running a marathon without a time goal is like driving without a destination — you'll get somewhere, but you won't know if you arrived where you meant to go.

Predicting your marathon finish time is one of the most useful things you can do before race day. A good prediction informs your training targets, your race-day pacing strategy, and your mindset at the start corral.

Start with a recent race result

The most reliable input for any marathon prediction is a recent race at a shorter distance — a 5K, 10K, or half marathon completed within the last three to six months, near your peak fitness.

Here's why: your recent race is the best observable evidence of your current fitness. A model that ignores your actual performance and relies solely on averages will always be less accurate than one anchored to a real effort you've produced.

The longer the reference race, the better. A half marathon result gives far more useful signal for marathon prediction than a 5K does — there's simply less distance to extrapolate across, and your endurance at race pace is directly observable.

The distance scaling problem

Running doesn't scale linearly. A runner who finishes a 5K in 20 minutes can't simply run 8.4 of them back-to-back for a marathon. Performance degrades with distance in a predictable — but non-trivial — way.

The classic prediction formulas capture this decay: they model how much slower humans run as distance increases. What they're really measuring is the point at which your aerobic and glycogen systems begin to fail.

This degradation rate varies from runner to runner, which is why any single formula applied universally produces errors. Some runners are speed-biased — they perform disproportionately well at shorter distances. Others are endurance-biased — they hold their pace unusually well over long efforts. If you have two data points (say, a 10K and a half marathon), a good prediction model can learn your personal fatigue curve rather than assuming you're average.

What else affects the prediction

Race time prediction isn't only about fitness. Several real-world factors make a meaningful difference:

Temperature. Running in heat is measurably slower. At 25°C, expect 5–8% more time compared to ideal conditions around 10–15°C. At 30°C or above, the penalty becomes significant enough to rebuild your entire pacing plan from scratch.

Elevation gain. Flat marathons and hilly ones are not comparable. Every significant climb costs time that descents only partially return. A course with 400m of gain will be several minutes slower than a flat one, all else equal.

Training load. A runner doing 70km per week who has completed a 32km long run is better prepared for a marathon than one doing 40km per week with a 20km long run — even if their recent 10K times are identical. The prediction should know this.

Speed work. Interval and tempo training improves your lactate threshold, which is especially important for maintaining pace in the second half of a marathon. It's one of the most consistent predictors of whether a runner will hold their splits or fade.

Why you need a range, not a single number

Most calculators give you a single number: "Your predicted marathon time is 3:42:17." That precision is false.

Marathon performance has genuine uncertainty — even for well-prepared runners. Your training may leave you slightly over or underprepared. Race-day conditions may differ from what you planned for. You might go out too fast (most runners do) or have a poor nutrition day.

A more honest prediction gives you a range. Something like: expect 3:38–3:52, most likely around 3:44. The range narrows when you have more data — two race results, full training details — and widens when you're predicting further from your reference race. A 5K to marathon prediction carries far more uncertainty than a half marathon to marathon one.

Single-number predictions encourage false confidence. Ranges prepare you for what race day actually looks like.

How to use a prediction on race day

Once you have a predicted finish time:

Set a target pace per kilometer. Divide the predicted time by the race distance. Build in a 3–5 second/km buffer for the first half — marathons are won or survived by going out conservatively.
Plan your fueling around your predicted duration. If you expect to be running for four hours or more, your glycogen strategy is different from a sub-3 runner's. Start fueling earlier than you think you need to.
Have a B target. If conditions on race day are worse than expected — a heat wave, unexpected hills, poor sleep — know in advance what time you'll shift to. Having a contingency plan reduces the mental cost of adjusting mid-race.
Trust the prediction through 30km. After that, it's guts, stubbornness, and whatever training left in your legs.

RaceCast takes your recent race result, optional training details, and race-day conditions — and runs three prediction models to produce a blended estimate with an honest confidence range. It takes about 30 seconds.