Wrist temperature is the most misread number on the Apple Watch, almost entirely because people expect a thermometer and get something else entirely. Once the framing is right, it becomes one of the more useful overnight signals.
Deviation, not absolute
The watch samples skin temperature at your wrist through the night, establishes what is typical for you over about five nights, and then reports each subsequent night as a signed departure from that personal baseline. Plus 0.4. Minus 0.2.
There is a good reason for this design. Absolute skin temperature at the wrist is close to useless as a health measure: it is dominated by ambient temperature, bedding, blood flow to the extremities, and how tightly your band is fitted. What is not useless is the change in that figure for one person across nights under broadly similar conditions.
So the number is small, signed, and only meaningful relative to your own history. Comparing your plus 0.3 with someone else’s plus 0.3 is meaningless.
What actually moves it
Your bedroom. The single largest confounder. A warmer room, a heavier duvet or a partner sharing the bed all raise it, and none of that says anything about your physiology.
Alcohol. Peripheral vasodilation raises skin temperature reliably. It is one of the reasons a drinking night moves almost every overnight metric at once.
The menstrual cycle. The most substantial genuine physiological driver for those who have one. Temperature rises after ovulation and stays raised through the luteal phase, which is roughly half of every cycle. Apple uses exactly this to provide retrospective ovulation estimates.
Illness. Often raises it a night or two before symptoms, usually alongside a rising heart rate and respiratory rate. The combination is the signal rather than temperature alone.
Hard training. A modest rise the night after a demanding session.
Alcohol, heat and travel together explain the large majority of unexpected spikes, which is worth remembering before concluding anything about your immune system.
Reading it usefully
The mistake is treating a single elevated night as information. The baseline itself takes five nights to establish and remains noisy for considerably longer, and a single night sits inside normal variation most of the time.
What is worth attention is a rise of several tenths sustained across two or three nights, particularly when your resting heart rate and respiratory rate have moved in the same direction over the same period. That combination is one of the stronger patterns wrist hardware can produce.
Even then, the base rate problem applies. Most such patterns are not the start of an illness, because most nights are not. We put numbers on exactly how often that inference misfires in your resting heart rate knows first.
Where it appears
In the Health app under Wrist Temperature, in Cycle Tracking for retrospective ovulation estimates, and as one of the five metrics in the Vitals app, covered in the Apple Watch Vitals app explained.