For runners · free, no account needed
Heat and humidity running pace calculator
Hot, humid weather slows every runner — the question is how much to adjust. This calculator turns your goal and the conditions (temperature, humidity, and sun) into a pace range based on published research you can check below. It can also pull the live forecast for your race.
Your race
Worked example
How much slower will I run a marathon in 75 °F heat?
It depends on the humidity and the sun, not the air temperature alone. Here is what this calculator returns for a 4:00 marathon goal (9:09 /mi) at 75 °F under three skies — the same model, the same inputs the form above asks for:
Dry and overcast
Assumes dew point 50 °F (10 °C), overcast.
Heat load: 18.8 °C (66 °F) WBGT — high, estimated yellow-flag territory.
+2.4% to +5.3% slower
9:22–9:38 /mi (+13 to +29 s/mi) · finish 4:05:43–4:12:36
Humid and overcast
Assumes dew point 68 °F (20 °C), overcast.
Heat load: 22.5 °C (73 °F) WBGT — high, estimated yellow-flag territory.
+3.2% to +7.5% slower
9:27–9:50 /mi (+18 to +41 s/mi) · finish 4:07:39–4:17:57
Humid and full sun
Assumes dew point 68 °F (20 °C), full sun.
Heat load: 24.3 °C (76 °F) WBGT — very high, estimated red-flag territory.
+3.6% to +8.7% slower
9:29–9:57 /mi (+20 to +48 s/mi) · finish 4:08:37–4:20:46
Same 75 °F, three different answers — which is why the form asks for dew point and sky rather than air temperature alone. The heat load is an estimate, not a measurement, and the flag colors are estimates against published cut-points; your race sets its own flags. The ranges are model estimates from the studies cited below, not a prediction for you. Enter your own goal and conditions above for your race.
The research — and our modeling choices
The model's anchor coefficients trace to the studies below. The choices that connect them — solar fractions, range caps, and cohort widening — are Wyndo modeling decisions documented separately from the published findings:
Ely MR, Cheuvront SN, Roberts WO, Montain SJ (2007). Med Sci Sports Exerc 39(3):487-493
Marathon slowdown grows with WBGT (studied range ~5-25 °C) and is cohort-dependent: fastest runners ~1% per +5 °C WBGT, three-hour runners roughly 1.5% per +5 °C — our upper bound.
El Helou N, Tafflet M, Berthelot G, et al. (2012). PLoS ONE 7(5):e37407 (1.8M marathon results)
Performance-vs-temperature is an inverted U: optimum roughly 4-10 °C, cooler for faster cohorts, with accelerating slowdown away from it — our cohort optima and curve shape.
Mantzios K, Ioannou LG, Panagiotaki Z, et al. (2022). Med Sci Sports Exerc 54(1):153-161
Large multi-race analysis found materially smaller heat slowdowns than Ely (roughly half to two-thirds), with observed maxima near ~8% — our lower bound and reality check.
Vihma T (2010). Int J Biometeorol 54(3):297-306 (Stockholm marathon)
Independent confirmation that slower finishers lose more time per degree of heat than faster ones — supports the cohort ordering.
Riegel PS (1981). American Scientist 69(3):285-290
Endurance-time scaling t2 = t1·(d2/d1)^1.06 — used only to place a non-marathon goal time into the marathon-based cohort bands.
Yaglou CP, Minard D (1957). AMA Arch Ind Health 16(4):302-316
Defines WBGT = 0.7·T_wet-bulb + 0.2·T_globe + 0.1·T_air — the heat-load scale this model and the race flag systems use. Ely and Mantzios report in WBGT; El Helou reports in air temperature (see the method notes).
Stull R (2011). J Appl Meteorol Climatol 50(11):2267-2269
Wet-bulb temperature from air temperature and relative humidity (valid roughly -20…50 °C, RH 5-99%) — used by our simplified WBGT estimate.
Armstrong LE, Casa DJ, Millard-Stafford M, et al. (ACSM position stand) (2007). Med Sci Sports Exerc 39(3):556-572
Event-alert flag categories for road racing; black flag above ~28 °C WBGT — where this calculator stops giving pace numbers and defers to the race’s flag guidance.
How it works — and what it doesn’t know
Your goal is placed into a cohort by marathon-equivalent time (slower cohorts slow more per degree). Conditions become a heat-load estimate (WBGT). Above your cohort’s optimum, the slowdown range grows with heat: the lower bound follows the smaller published effect (Mantzios 2022), the upper bound the larger one (Ely 2007, accelerating per El Helou 2012’s curve). One documented approximation: El Helou reported optima in air temperature, not WBGT; on the cool mornings where those optima sit the two run close, so the model treats them as approximate WBGT values.
Not modeled in v1: wind (so treat the heat estimate as approximate in breezy conditions), heat acclimatization, course shade or surface, and cold-weather slowing. The WBGT estimate is a documented simplification; a full physical model (Liljegren) is the planned upgrade.
This is planning guidance from published field studies — not medical advice, and never a green light to race in dangerous heat. Above ~28 °C WBGT the calculator stops giving pace numbers and defers to your race’s event-alert flags.