Wrist temperature during sleep: what Apple Watch is actually tracking

If you wear an Apple Watch Series 8 or later, or an Apple Watch Ultra, to bed, it quietly logs a wrist-temperature reading every night — and most people who see it in the Health app have no clear idea what the number actually is, only that it moves. This article picks up where HRV explained leaves off: another overnight wearable signal that gets read as a verdict on the day when it was really only ever built to be read as a trend against your own history. A note before we start: this is informational material, not medical advice. Wrist temperature is a wellness signal, not a diagnostic test, and it isn't built to answer "am I sick" or "did I ovulate" on its own, on any single night.

What the sensor reports, and what it doesn't

The most common misunderstanding is the simplest one to correct: Apple Watch does not report your body temperature. Series 8 and later, and all Apple Watch Ultra models, carry two temperature sensors — one on the back crystal near the skin, one just under the display — and while you sleep, the watch samples both every five seconds, aggregating thousands of readings into a single nightly figure designed specifically to reduce bias from the surrounding environment (Apple Support, wrist temperature). That figure isn't an absolute reading like "36.7°C." It's a deviation from a personal baseline that the watch establishes from your own data after about five nights of sleep tracked for at least four hours with Sleep Focus on — something like "+0.2°C" or "−0.4°C" relative to what's typical for you, not relative to any population average or clinical norm. Apple's own documentation is direct about the limits of this: the feature is not a thermometer, can't be checked on demand the way a pulse oximeter reading can, and — its own words — "is not a medical device, and is not intended for use in medical diagnostic purposes" (Apple Support, wrist temperature). Two consequences follow directly from that design. First, the number is retrospective by construction: it's a nightly average computed after the fact, not a live reading, so there's no version of "check your wrist temperature right now" the way there is for heart rate. Second, because it's relative to your own history rather than absolute, a reading only becomes interpretable once your watch has enough of your own nights behind it to know what's typical for you — exactly the same baseline-first logic the HRV article covered for a different overnight signal.

Why temperature moves at all around sleep

Before getting to what a shift from baseline might reflect, it's worth understanding why wrist and skin temperature move overnight in the first place, because the mechanism is a genuinely well-established piece of sleep physiology, not a wearable-specific quirk. A body preparing to fall asleep redistributes heat from its core toward its periphery — hands, feet, and wrists dilate their blood vessels and warm up, while core temperature drops slightly, and this heat redistribution is tightly coupled to how quickly someone actually falls asleep. A meta-analysis of studies conducted under controlled laboratory conditions found that this heat-loss signal — measured as the gap between proximal (core-adjacent) and distal (peripheral) skin temperature — was a better predictor of sleep-onset latency than core body temperature itself, heart rate, melatonin onset, or how sleepy someone said they felt (Kräuchi, 2001). In plain terms: your wrist warming up isn't an incidental side effect of sleep, it's part of the physiological process of falling asleep, and it's one of the more reliable windows researchers have found into how ready the body actually is for sleep at a given moment, ahead of the sleep itself. That's useful context for reading an Apple Watch chart, because it means the nightly number reflects a real, mechanistic process tied to your own sleep timing and depth — not just an incidental artifact of a warm blanket, though a warm blanket can shift it too, which is the subject of a later section.

The cycle: a biphasic shift as an observed pattern

The single best-characterized reason wrist temperature moves over weeks rather than nights is the menstrual cycle. Progesterone, released after ovulation, produces a small, sustained rise in temperature — a biphasic shift, cooler in the first half of the cycle and slightly warmer for most of the second — and this pattern is well established enough that Apple Watch Series 8 and later, and Apple Watch Ultra, are built specifically to detect it: using your logged cycle data alongside nightly wrist-temperature readings, the watch looks for that shift and, once detected, retrospectively flags an estimated day of ovulation, typically appearing as an estimate a few days after the fact rather than in real time (Apple Support, ovulation estimates). This isn't a marketing flourish layered on top of a wellness gimmick — it's now backed by a 2025 prospective cohort study that followed roughly 260 women through close to 900 menstrual cycles, evaluating exactly this kind of wrist-temperature algorithm. In cycles where the wrist sensor picked up a clear signal (a change of at least 0.2°C, the threshold the algorithm uses to flag ovulation-linked warming), the same temperature data also carried useful information about timing further out: an estimate of the next cycle's start date landed within three days of the actual start about 89% of the time, with a typical error of roughly 1.65 days (Wang et al., 2025). That's a genuinely useful, independently studied result — and it's still, by the researchers' own framing and Apple's, an estimate built from an observed pattern relative to your own baseline, not a real-time readout and not a substitute for tracking your actual cycle if you're relying on timing for anything that matters.

Feeling unwell: a shift, not a diagnostic marker

The same physiology that makes wrist temperature useful for cycle tracking gives it a second, less-marketed use: as one of the earliest observable signals that something ordinary might be off, days before it announces itself another way. Fever is, mechanistically, the body deliberately raising its core and peripheral temperature set point in response to infection, and continuous wearable temperature monitoring has actually been tested directly against this. A 2020 study using a wearable device that continuously tracks peripheral temperature followed 50 people who went on to report COVID-19 infections and found that illness-associated elevations in peripheral temperature were observable in the wearable data and tracked with participants' own reports of fever — in many cases with a detectable temperature rise appearing before the person felt clearly unwell (Smarr et al., 2020). That's a real, if modest and single-condition, demonstration that a sustained shift away from your nightly baseline can be an early, honest signal that something is different — not that it is any specific illness, and not a diagnostic marker in the way a clinical thermometer reading against a fixed threshold is. The distinction matters: a fixed threshold like "38°C is a fever" only works because it's compared to a known, population-wide normal core temperature. Apple Watch's wrist sensor has no such fixed reference — it only ever tells you that tonight looks different from your own recent nights, which is a genuinely useful thing to notice and a genuinely different claim from "you have a fever," let alone what's causing one.

What adds noise to your own baseline

Because the number is a deviation from your own recent nights rather than a fixed population reference, anything that changes your sleep environment or physiology from one night to the next can move it independent of cycle phase or illness — and Apple's own guidance is explicit that this is expected, not a malfunction: "your body temperature naturally fluctuates, and can vary from night to night due to a number of variables," including your sleep environment itself (Apple Support, ovulation estimates). A warmer or cooler bedroom than usual, a heavier or lighter blanket, a late intense workout, alcohol, or simply a night with unusually little or fragmented sleep tracked by the watch are all ordinary, explainable reasons a single night's reading could sit away from your baseline without reflecting your cycle or your health at all — the same category of everyday noise the HRV article described for overnight heart-rate variability, on a different signal with the same underlying lesson: a wearable reading is downstream of everything happening in a given night, not just the one thing you're hoping to learn about.

What to do with the number, practically

Given how the signal is built — retrospective, relative to your own history, and sensitive to ordinary nightly noise — the useful habit is the same one this whole cluster keeps landing on: read it as a multi-night pattern next to your own context, not a verdict from a single night. A one-night jump after a late workout or a warmer-than-usual room is exactly the kind of noise described above and isn't, on its own, informative about anything. A sustained shift held across several consistent nights, especially one that lines up with where you are in your cycle or with actually feeling off, is the pattern worth paying attention to — and that's precisely what MeteoHealth's Sleep section is built to make visible: it reads the sleep and wrist-temperature data your Apple Watch already collects through Apple Health and shows it against your own baseline, alongside your logged symptoms and the local weather, on-device, so a shift in the number sits next to the context that might actually explain it rather than floating on its own as an unexplained number on a screen. It doesn't diagnose a cycle, an illness, or a sleep disorder, and it doesn't tell you what to do about any of them — it shows you your own data, next to itself, over time.

Bottom line

Wrist temperature on Apple Watch is a real, physiologically grounded signal — tied to the same core-to-periphery heat redistribution that helps drive sleep onset, sensitive enough to pick up the small, well-documented biphasic shift of the menstrual cycle, and sensitive enough, in at least one published study, to register illness-associated warming before people felt clearly unwell. But every one of those uses depends on the same design choice: the number is a deviation from your own recent nights, computed after the fact, not a live thermometer reading against any universal scale. A warm room, a late glass of wine, or one restless night can move it exactly as much as a real physiological shift can, which is why the only honest way to use it is the one this whole content series keeps repeating: watch the pattern against your own baseline, over enough nights to tell a signal from noise, and read any single number as a starting point for noticing — not a diagnostic answer on its own.