Every number on this page describes a notional adult: healthy, dressed for the season, walking rather than working. It is a weather index, not a physiological one, and the gap between those two things is where people get hurt.
You. Age, fitness, acclimatisation, body size, medication and clothing all move the real answer more than a few km/h of wind does. Someone who has spent a fortnight in the heat tolerates it markedly better than someone who flew in yesterday, and no term in any of these formulas knows that.
WBGT. Wet-bulb globe temperature is what sports governing bodies and workplace heat regulations actually use, because it measures radiant load and air movement directly rather than inferring them. If you need a threshold to make a decision against, it is WBGT you want, not heat index.
The assumptions inside each formula. Wind chill assumes an exposed adult face at 1.5 m in the shade, walking at 1.4 m/s, and wind measured at the standard 10 m — which is roughly 1.5× what you feel at head height. Heat index assumes shade and a light breeze. Apparent temperature assumes shade unless the sun toggle is on.
The far ends. Below about −50 °C and above about 50 °C all three are extrapolating past the data they were fitted to. The numbers keep coming; their meaning does not.
Wind chill and heat index are checked against the published NWS charts in feelslike_model.py — seven wind-chill cells and six heat-index cells, all reproduced exactly. That model also caught two things in this page: the last Rothfusz coefficient had been typed as 1.99788×10⁻⁶ where the regression publishes 1.99×10⁻⁶, worth up to 1.1 °F at the extreme corner, and the sun toggle was adding a flat +15 °F to the heat index rather than modelling radiation.
Below about 10°C: the NWS / Environment Canada 2001 formula, how much faster moving air strips heat from exposed skin.
Above about 27°C: the NWS Rothfusz regression, how humidity blocks sweat from evaporating, the body’s main cooling mechanism.
Between them an apparent-temperature blend, because wind and humidity still matter there. Each formula fades into the next so the line never jumps.
NWS bands of 30, 10 and 5 minutes to frostbite on exposed skin. Clothing, wetness and exertion still dominate real risk.
Wind chill stops at 10 °C. Heat index does not begin until 27. Between them sits most of the British year, and until now this page reported that wind and humidity did nothing there. They do.
“Feels like −13” does not mean the air is at −13: exposed skin loses heat as fast as it would in still air at −13. Your kettle will not freeze any quicker, pipes and car engines do not care, and a thermometer in the wind still reads the air temperature. Only living things feel wind chill. It can never cool skin below the air temperature, so frostbite from cold alone is not possible above 0°C.
It stops you cooling. Sweat works by evaporating, and evaporation slows as the air fills with water. At 35 °C and 20% humidity you are fine; at 35 and 80% you are in danger: same air, same thermometer. That is why the heat index climbs so steeply up the humidity axis and so gently along the temperature one. NWS testing found humidity changes wind chill by under a degree, which is why that formula leaves it out.
You enter the weather-station wind, measured at 10 m. The NWS formulas use wind at about 1.5 m, face height for a standing adult, and adjust for it: a forecast of 30 km/h is nearer 20 at head height in the open, and far less between buildings. If you enter what a weather app told you, that is the right figure to use.
The official chart is defined for shady, light-wind conditions. Direct sun can add up to 15°F (8°C); toggle direct sun exposure and the calculator applies the NWS full-sun adjustment. There is no equivalent for wind chill, which is about heat loss to moving air, not solar gain.
Estimated from temperature and relative humidity with the Magnus–Tetens approximation. High dew points (roughly above 18°C / 65°F) feel muggy; low ones feel dry. Separate from wind chill and heat index.
Three formulas cover three ranges, and they disagree where they meet: switching made the number jump by up to 5°C at 10°C and 4.8°C at 27°C. Each now fades into the next across the few degrees at its edge, weighted so the slope is continuous as well as the value. Likewise the NWS wind-chill formula is undefined below 3 mph, so the tool tapers smoothly from no effect at 0 mph to the formula’s own value at 3: a display smoothing choice on top of the formula, not a change to it.
NWS testing found humidity changed the result by less than a degree, so the official 2001 formula leaves it out entirely to keep the calculation simple and reliable — wind chill is driven almost entirely by temperature and wind speed.
No. Wind chill can never cool human skin below the actual air temperature — it only describes how much faster heat is lost. If the air temperature itself is above 0°C, frostbite from cold exposure alone isn't possible, whatever the wind chill number says.
The core Rothfusz regression is a statistical fit, not a physical law, and it drifts at the extremes — the NWS adds small correction terms for very low humidity (which lowers the index, since sweat evaporates efficiently) and very high humidity at moderate heat (which raises it further).
For heat index, yes, and it's built in above — toggle "direct sun exposure" and this calculator adds the NWS full-sun adjustment (up to 15°F / 8°C over the shaded value) directly. There's no equivalent sun adjustment for wind chill, since the formula is about heat loss to moving air, not solar gain.
Because three formulas cover three ranges, and only one of them is defined everywhere. Below 10°C the answer follows the NWS wind chill; above 27°C, the NWS heat index; in between, the apparent temperature — which is not simply the air temperature, because wind and humidity go on mattering through the mild months. The three disagree where they meet, so switching between them made the number jump: up to 5°C at the 10°C boundary and 4.8°C at 27°C. Each now fades into the next across the few degrees at its edge, weighted so the slope is continuous as well as the value. The line changes gradient around 10°C and 27°C; it no longer breaks there.
The NWS formula is only defined for wind above 3 mph — below that, it doesn't publish a value at all, it just reports the air temperature unchanged. Taken literally that's a step: drag the wind slider from 2.9 to 3.1 mph and the number would jump several degrees in an instant, which doesn't match how wind actually behaves. This calculator tapers smoothly from no effect at 0 mph up to the formula's own value at 3 mph and beyond, so the number always moves continuously as you drag the slider — a display smoothing choice on top of the formula, not a change to it.
The temperature at which air would become saturated (100% RH) if cooled at constant moisture content. This tool estimates it with the Magnus–Tetens approximation from air temperature and relative humidity — useful comfort context, separate from the NWS wind-chill and heat-index formulas.
Releases in which this page changed, newest last. Derived from the archived copy of every release, not from notes written afterwards — so it reflects what actually shipped. Site-wide passes are left out; they are in the full changelog.