Wrist temperature on Apple Watch: what the number actually measures

Short answer

Your watch does not show your temperature. It shows how far last night sat from your own baseline, as a plus or minus — because a wrist is not a reliable place to measure absolute body temperature. It needs about five nights before it shows anything at all, and it is not a thermometer or a fever detector.

If you have looked at "+0.4" and had no idea whether that is good, bad or meaningless, that is not your fault and you did not miss something obvious. The number is deliberately relative, and almost nothing on the screen explains what it is relative to.

Note: this article explains a wellness feature and is not medical advice. Temperature sensing on consumer wearables is not a medical device and is not intended for diagnosis. If you think you have a fever, use a thermometer.

What the watch actually measures

Two sensors, not one, and the second is the interesting part.

Apple Watch Series 8 and later, all Ultra models and the SE 3 carry one sensor on the back crystal, sitting against your skin, and a second just under the display, facing outwards. The outer sensor exists to estimate what the room is doing, so the software can subtract the environment from the reading.

That second sensor is there because a wrist is a bad place to take a temperature. It is exposed to the air, to bedding, to whether your arm is under the duvet or outside it, to a draught, to how warm the room is. A finger sensor in a ring has some of these problems too, but a wrist has them worst.

It only measures while you sleep. The watch samples every few seconds through the night and averages those readings into a single nightly value. Daytime is far too noisy — your arm moves, sleeves come on and off, you go outside — so no daytime figure is offered at all.

Why it shows a deviation instead of a temperature

This is the design decision that confuses everyone, and it is the right one.

An absolute skin temperature from a wrist would be close to meaningless. Skin is not core body temperature, wrists run cooler than mouths, and the figure would shift with the room. Apple's answer is to stop pretending: learn what is normal for you, then report departures from it.

Hence the five-night wait before anything shows. Until the baseline exists the chart says "Needs More Data" and counts down remaining nights, which reads like a bug and is not one. It is also why the feature stays uninformative if you wear the watch at night only occasionally — the baseline never settles.

The practical consequence: your number is not comparable to anyone else's. A friend's +0.3 and your +0.3 are measured against different baselines and mean different things. This is the same logic behind a good HRV range and every recovery score — personal baseline, not population norm.

What moves the number

More things than people expect, which is exactly why a single night says little.

The menstrual cycle. The largest regular driver in anyone who has one. Progesterone after ovulation raises body temperature by roughly 0.2 to 0.5°C, and it stays raised until the period starts. If you are looking at a chart that steps up halfway through the month and drops back, you are looking at your cycle rather than anything going wrong.

Alcohol. Reliably raises overnight skin temperature, usually along with a higher resting heart rate and suppressed heart rate variability. This trio moving together is one of the most consistent patterns in wearable data.

Illness. Often the earliest thing to shift, sometimes a day or more before symptoms. This is the use people most want, and the caveats below matter.

The room, the bedding and your arm. Warm room, heavy duvet, arm under the covers. The outer sensor corrects for some of this and not all of it.

Exercise late in the evening, a hot bath, a heavy meal, and for some people hot flushes in perimenopause.

Fit. A loose band lets air in and produces noise that looks like biology.

Sleeping position. The arm you lie on runs warmer than the one in the air, and people who habitually change sides see more variation than people who do not. This is one of the more common explanations for a chart that looks erratic without any lifestyle change behind it.

Perimenopause and menopause. Hot flushes at night raise overnight readings, and the loss of the regular luteal rise changes the pattern the chart has been showing for years — which can read as something going wrong when it is something ordinary.

Shift work and travel across time zones. Body temperature follows a circadian rhythm, so measuring during a displaced sleep window measures a different point on that curve. Several nights after a long flight will look unusual for reasons that have nothing to do with health.

Can Apple Watch detect a fever?

No, and it is not designed to. The feature reports a deviation from your sleeping baseline, not a body temperature, and Apple states plainly that it is not a medical device and not intended for diagnosis.

What it can sometimes do is show a rise before you notice feeling ill. That is genuinely interesting and it is not the same as detecting fever. Three reasons not to lean on it:

Skin is not core. Wrist skin temperature and the number a thermometer gives you are different measurements that move together loosely.

A deviation has no threshold. There is no wrist equivalent of 38°C. Your +0.8 and someone else's +0.8 are not comparable, and no manufacturer publishes a cut-off, because there isn't a defensible one.

It is non-specific. Alcohol, a warm room, the luteal phase and an infection all push the same number the same direction.

So the honest use is narrow: if the number is up, and your resting heart rate is up, and your heart rate variability is down, that combination is worth taking seriously as a reason to rest and watch. If you actually want to know whether you have a fever, use a thermometer.

Wrist temperature and ovulation

This is the most developed application of the sensor, and it is worth being precise about what it offers.

If Cycle Tracking is enabled, the watch uses overnight temperature to provide retrospective ovulation estimates — that is, to tell you afterwards when ovulation probably happened, and to improve future period predictions.

The physiology is well established. Progesterone after ovulation produces a sustained temperature rise of about 0.2 to 0.5°C that persists until menstruation. That is the basis of traditional basal body temperature charting, done for decades with a thermometer first thing in the morning; the watch automates the measurement and removes the need to take it at the same moment every day.

Retrospective is the operative word. The shift appears after ovulation, so this confirms rather than predicts. It is useful for understanding your own pattern over months. It is not contraception, and it is not a fertility treatment tool — anyone trying to conceive or avoid conception on timing alone needs a method built for that purpose and a clinician's input.

One genuinely useful thing it can surface over several months: cycles where the expected rise does not appear at all. That pattern has possible explanations worth asking about rather than interpreting yourself.

Why illness sometimes shows up before you feel it

This is the claim that sells temperature sensing, and it is half true in a way worth unpacking.

The body raises its temperature as part of mounting an immune response, and that process begins before you consciously register feeling unwell. So a rise appearing the night before a sore throat is not magic — it is a physiological event preceding a perceptual one.

But a rise is not a prediction. It is an early observation of something already happening. The distinction matters because it changes what the information is good for: not avoiding the illness, but deciding earlier to cancel things, rest, and not infect a room full of people.

And most raised nights are not illness. This is the part that gets lost. Over a year, far more of your elevated readings will come from alcohol, a warm bedroom, a late meal and the second half of your cycle than from infection. Treating every rise as an early warning produces a lot of false alarms and a fair amount of anxiety.

The combination is what carries signal. Temperature up, resting heart rate up, heart rate variability down, all three together, across more than one night — that shape is much more specific than any one of them. It is also the shape that appears in the two or three days before many people notice symptoms, and it is what makes an early-warning reading worth anything at all.

What to do with it is modest. Sleep more, drink water, drop the optional things from tomorrow, and see. That is the entire correct response. It is not a reason to take medication, and it is not a reason to test for anything in the absence of symptoms.

How accurate is it?

Two different questions live inside that one, and separating them resolves most of the confusion.

Is the absolute value accurate? Wrong question. The watch does not claim to give you a body temperature and does not display one.

Is the deviation reliable? Reasonably, for overnight measurement, with conditions. It needs consistent wear, a decent fit and a settled baseline. The two-sensor design handles some environmental noise and not all of it. And the whole thing rests on sleeping in the watch — which is where most of the real-world failure happens, because a watch on a charger measures nothing.

The structural comparison worth knowing: a finger runs warmer and more stably than a wrist, which is why ring-based devices have an easier job with this particular signal. That is a hardware advantage, not a software one, and it is the same reason the finger has an edge for overnight HRV.

The more useful accuracy question is not about the sensor at all. It is how many nights you have. A deviation is only meaningful against a stable baseline, and a baseline built from patchy wear is not stable. Six weeks of consistent nights beats a better sensor worn half the time.

How to read your own chart

Look at multi-night stretches, not single nights. One elevated night is noise. Three consecutive nights up is a pattern.

Know where you are in your cycle before reading anything into a rise, if you have one. The luteal phase explains a large share of what people find alarming.

Read it alongside the other two. Temperature up, resting heart rate up and heart rate variability down, moving together, is a far stronger signal than any of the three alone — and it is the classic shape of an infection arriving.

Account for the obvious first. Alcohol, a hot room, a late workout, a heavy meal. These explain most single-night spikes.

Give a new watch six weeks before concluding anything about your pattern.

And do not chase the number. There is no target. Lower is not better, and a flat line is not a health achievement — it is a measurement of your bedroom as much as of you.

What the number cannot see

Worth listing, because the gaps are where a confident-looking figure misleads.

Your actual body temperature. It is skin at the wrist during sleep, and the relationship to core temperature is loose. This is the whole reason the display is a deviation.

Anything during the day. Sampling stops when you stop sleeping, so a fever that arrives at two in the afternoon does not appear until that night — if it still exists.

What caused the change. The single most important limitation. A raised night from an infection and a raised night from a glass of wine are identical on the chart.

Medication. Some drugs affect thermoregulation, and hormonal contraception flattens the cyclical pattern the feature partly relies on. Neither is visible to the watch, and both change what your chart means.

Nights you did not wear it. Obvious and easy to forget: gaps do not just lose data, they degrade the baseline that every future reading is measured against.

How to bring this up with your doctor

The reading is rarely the reason to go. It can be what makes you notice a pattern worth describing.

Take a real temperature first if you feel unwell. A thermometer answers the question the watch cannot, and it is the number a clinician can use.

Bring the trend and its companions, not a screenshot. "My overnight temperature has run above my usual for three weeks, with a higher resting heart rate and lower HRV" is a usable sentence. "My watch says plus 0.6" is not, because nobody else knows your baseline.

Say what you have ruled out. Alcohol, room temperature, where you are in your cycle. Having excluded the obvious makes the rest credible.

Ask for the ordinary things to be checked if the pattern has persisted: full blood count, inflammatory markers, thyroid function. Persistent low-grade temperature elevation has a real differential and it is worth working through.

If the luteal rise is repeatedly absent across several tracked cycles, raise that specifically rather than interpreting it yourself — it has explanations worth asking about.

And do not let the watch become the conversation. The useful framing is your symptoms and their timeline, with the data as supporting evidence rather than as the complaint.

When to see a doctor

The temperature reading is not the reason to go, but it can be the thing that makes you notice a pattern worth mentioning.

Take a real temperature first if you feel unwell. A thermometer answers the question the watch cannot.

Persistent unexplained fever — days of raised temperature without an obvious infection — needs assessment, and that is true whatever your watch says.

Night sweats with weight loss or persistent fatigue are a specific combination that should be looked at, and not one to monitor at home.

If the luteal rise is repeatedly absent across several cycles while you are tracking, that is worth raising with a clinician rather than interpreting yourself.

Bring the trend, not the screenshot. "My overnight temperature has been running above my usual for three weeks, along with a higher resting heart rate" is a useful sentence. "My watch says plus 0.6" is not, because your doctor has no idea what your baseline is.

How Welltory helps — and what it cannot do

The limit first. Welltory does not measure temperature, does not detect fever and does not diagnose anything. It is a general wellness product with no regulatory clearance.

What it does is work with the rest of the picture, which is where temperature becomes interpretable. On its own a deviation is ambiguous; alongside resting heart rate and heart rate variability read against your own months of history, it starts to mean something. Welltory reads those from Apple Health or a connected wearable, or takes a reading through the phone camera if you have no device at all.

The principle is the same one that makes the watch show a deviation rather than a number: a measurement means something against your own baseline and very little against a chart. That is why the useful comparison is always you in March against you in September.

Two honest caveats. All of these signals are non-specific — they move with illness, alcohol, heat, stress and poor sleep alike, so a change says something is different, never what. And none of it replaces a thermometer, a clinician, or your own judgement about whether you feel ill.

How we made it

Made with AI tools, then edited and fact-checked by the Welltory team. See our Editorial & AI policy.

Data analysis by Jane Smorodnikova, co-founder of Welltory and the person who built the methodology behind how we read physiological data.

Written by Tatsiana Yashyna.

Discounts for blog readers: up to 36% off

See what affects your energy, stress, sleep, and daily state with Welltory

Explore your data