100°F to °C

100 degrees Fahrenheit is 37.78 degrees Celsius

One hundred degrees Fahrenheit is 37.78 degrees Celsius. The exact value runs 37.7778 with the 7 repeating, because the arithmetic lands on the fraction 340 / 9: 100 - 32 = 68, then 68 x 5 = 340, then 340 / 9 = 37.7778. Rounded to one decimal, which is what weather apps and recipe cards use, the answer is 37.8 °C.

The repeating part is not a rounding accident. Fahrenheit and Celsius degrees sit in a ratio of 9 to 5, and 68 does not divide evenly by 9, so the decimal cycles through 7s forever. Write the result as a fraction, 37 and 7/9 °C, and nothing is lost. Write it as 37.78 and the error is 0.002 °C, smaller than any home thermometer can resolve.

The neighbors of 100 are worth keeping on hand, because forecasts and fever readings cluster here:

Read the table as a ratio rather than a list: every 2 Fahrenheit degrees is 1.11 Celsius degrees, so each row climbs by the same amount. That constant climb is why conversion feels mechanical once you run it a few times, and why 100 needs a decimal while 212 °F lands on a clean 100 °C.

The formula, worked forward and backward

NIST Special Publication 811, the institute's guide for using SI units, gives the conversion as t/°C = (t/°F - 32) / 1.8 and marks the 1.8 as an exact factor (NIST, 2008 edition). Applied here: 100 - 32 = 68, and 68 / 1.8 = 37.7778 °C. The fraction form, (100 - 32) x 5/9, produces the same number, because dividing by 1.8 and multiplying by 5/9 are the same operation.

Running it backward is the fastest way to catch a mistake. Start from 37.7778: 37.7778 x 1.8 = 68.0000, and 68 + 32 = 100. When a round trip does not land back on the starting value, a step went in the wrong order. The two usual order errors are subtracting 32 after multiplying, and adding 32 to a temperature difference instead of to a reading.

The degree size matters as much as the offset. NIST lists the Fahrenheit to Celsius interval factor as 5.555556 E-01, so each single Fahrenheit degree is 0.556 Celsius degrees, and each Celsius degree is 1.8 Fahrenheit degrees. That ratio is the whole conversion; the 32 only decides where the counting starts. The NIST conversion factor tables carry both entries, and both constants are exact by definition.

Rounding 37.7778: when 37.8 and 38 are different answers

Three roundings of this value circulate, and they are not interchangeable. 37.78 keeps the hundredths and discards 0.002 °C. 37.8 keeps one decimal and discards 0.022 °C. Writing 38 °C throws away 0.22 °C, which is 0.4 Fahrenheit degrees, and that is large enough to cross a clinical boundary.

The trap works in both directions. A Celsius report of 38.0 °C converts to exactly 100.4 °F, because 38 x 1.8 = 68.4, and 68.4 + 32 = 100.4. So the round Celsius number 38 and the round Fahrenheit number 100 sit 0.4 °F apart, even though both read as "the same temperature, rounded". Which one a report publishes changes what the reader concludes.

A practical rule: keep two decimals for anything medical or scientific, one decimal for weather and cooking, and whole degrees only for small talk. 37.78, 37.8 and "about 38" are all defensible answers to the same question, as long as the context says which precision is in play.

A thermometer reading 100 °F: fever or not

The CDC's reporting definitions set the fever line at a measured temperature of 100.4 °F, or 38 °C, for travelers and crews (CDC port health guidance). Mayo Clinic guidance treats an oral reading of 100 °F, or 37.8 °C, as generally a fever. A thermometer showing 100 °F, which is 37.78 °C, therefore sits 0.4 °F below the CDC threshold and right at the Mayo Clinic line. Whether it counts as a fever depends on which guidance and which measurement site is in play.

It helps to see the round numbers for what they are. 98.6 °F is the exact Fahrenheit rendering of 37 °C: 37 x 1.8 = 66.6, and 66.6 + 32 = 98.6. Likewise 100.4 °F is 38 °C to the decimal. The conventional body average and the CDC cutoff are both round Celsius values wearing Fahrenheit clothes, while 100 °F is a round Fahrenheit value that lands on the untidy 37.7778.

The table shows how tightly the landmarks pack together here. The gap from the traditional average to the CDC line is 100.4 - 98.6 = 1.8 °F, exactly one Celsius degree. There is no setting where 100 °F is unremarkable: it stands 1.4 °F above the traditional average and 0.4 °F below the CDC line.

100 °F outdoors: the heat index runs from 96 to 124

Air temperature alone understates a 100 °F day. The National Weather Service combines temperature with relative humidity into the heat index, and its chart puts 100 °F at 15 percent humidity at a feels-like 96 °F, while the same air temperature at 55 percent humidity reads 124 °F. Full sunshine can push the number up by as much as 15 °F more (NWS heat index guidance). The spread between those two examples is 28 °F, and all of it comes from humidity.

In Celsius-speaking countries the same afternoon appears on the forecast as 37.8, and heat warnings there key off similar numbers. The conversion does not change how the day feels; it changes the label. What matters for planning is the heat index band, not the air temperature alone: a 100 °F reading in dry mountain air and a 100 °F reading on a humid Gulf Coast afternoon are different days carrying the same number.

Water, pools and ovens at 100 °F

Water at 100 °F is 37.78 °C, which lands just above the traditional body average of 37 °C and just below the 38 °C fever line. A pool held there feels like bathwater: pleasant for soaking, too warm for exercise, and nothing like the high 70s to low 80s in Fahrenheit where lap pools are usually kept. Tap water at 100 °F is a warm shower, not a scald.

In the kitchen, 100 °F is below every baking and roasting temperature on a recipe card. It appears only in the gentler jobs: warming plates, holding food briefly, or the bread proof setting some ovens carry, which aims near this range. Candy work and frying live hundreds of degrees higher, and the 100 mark on a dial is not a setting any baking method uses.

Laboratory and aquarium work give the number a third life. Converted to 37.78 °C, the reading drifts in ways that matter at small scales: an error of 1 °F is 0.56 °C, which is large for an incubator or a reef tank and invisible in a hot tub. Precision there comes from the same formula, only with tighter rounding.

The halve after 30 shortcut, and where it drifts

The mental method most people learn runs: subtract 30, then halve. Applied here, (100 - 30) / 2 = 35, against the true 37.78, so the estimate lands 2.78 degrees low. The shortcut was built for the middle of the weather range, where 50 °F gives (50 - 30) / 2 = 10 and the true value is 10.00 °C, an exact hit. Away from 50 it drifts downward, and the drift grows with the reading.

Running the shortcut the other way is worse at this end. Doubling 37.8 and adding 30 gives 105.6 °F, which is 5.6 degrees above the true 100. When the job needs better than 3 degrees of accuracy, do the real arithmetic in two steps: subtract 32, then take 0.556 of the result. On 100 that is 68 x 0.556 = 37.8, within 0.03 of the exact 37.7778 in a single move.

Kelvin, Rankine and the one-line summary

The two remaining scales are plain offsets from the Celsius answer. Kelvin: 37.7778 + 273.15 = 310.93 K. NIST gives the direct route as T/K = (t/°F + 459.67) / 1.8, and 559.67 / 1.8 = 310.93, the same place by a different road. Rankine shares a degree size with Fahrenheit, so it takes one offset: 100 + 459.67 = 559.67 °R.

Everything on this page compresses into one line: 100 °F = 37.7778 °C, which rounds to 37.8, and equals 310.93 K and 559.67 °R. Each Fahrenheit degree away from here moves the Celsius value by 0.556, so 101 °F is 38.33 °C and 99 °F is 37.22 °C. The widget above computes any of these from a number you type, including the heat index at a humidity you set.

State the precision when you pass the number along. A weather report, a doctor's office and a lab notebook each want a different rounding of the same 100 °F, and the difference between 37.78 and 38 can be the difference between a reading inside a normal range and one outside it.