Apple shipped Sleep Score in watchOS 26, years after Oura, WHOOP and Fitbit had one. What arrived is narrower than any of those, and the narrowness is the interesting part. Once you know the three weights you can look at any night and say precisely why it scored what it scored, which is not true of most competing scores.
The three components, and what each is really measuring
Sleep duration, up to 50 points
Half the score. This is time actually asleep, not time in bed, compared against the sleep goal you set in the Health app. The distinction matters more than people expect: if you are in bed for eight hours and asleep for six and a half, this component scores the six and a half.
Because it is weighted at 50, duration dominates everything else. An extra 45 minutes of sleep will move your score further than any amount of tidying up your bedtime routine.
Bedtime consistency, up to 30 points
This looks at when you fell asleep across the previous 13 nights, and rewards a narrow spread. It is the component people find least intuitive, because it is the one that penalises you for something that happened last week.
The practical consequence: a single 2am Friday does not just cost you that night, it widens your consistency spread and shaves points off every score for the next fortnight. If your score seems stuck in the seventies despite good nights, this is almost always the reason.
Interruptions, up to 20 points
Both how often you woke and how long you stayed awake. Brief stirrings that you do not remember are normal and cost little. A 40-minute stretch staring at the ceiling at 3am costs real points.
What the score deliberately ignores
Sleep stages. Your watch estimates REM, Core and Deep every night and shows them in a chart directly above the score. None of them are inputs. You can have a night that is unusually light on deep sleep and still score in the nineties.
This is defensible. Consumer sleep staging from a wrist accelerometer and an optical heart sensor is the least reliable thing these devices produce, and building a headline score on it would be building on sand. Apple appears to have made exactly that call.
Physiology. Heart rate, HRV, respiratory rate and wrist temperature are all recorded overnight and none of them touch the Sleep Score. They live in the Vitals app instead, which is a separate feature with a separate job.
So the Sleep Score answers “did you sleep enough, regularly, and without interruption?” It does not answer “did your body recover?” Those are different questions and it is a point in Apple’s favour that it does not pretend otherwise.
The bands, and the change in watchOS 26.2
Apple sorts the 0 to 100 number into five labels: Very Low from 0 to 40, Low from 41 to 60, OK from 61 to 80, High from 81 to 95 and Very High from 96 to 100.
Those ranges were revised in watchOS 26.2. The score computation did not change, but the label attached to a given number did, which produced a wave of people convinced their sleep had deteriorated after an update. If that was you, it almost certainly was not.
The general lesson holds for every scored wearable metric: the number is more stable than the vendor’s adjectives, and your own trend is more informative than either.
How it compares to the other sleep scores
Oura and WHOOP both fold physiology into their sleep and readiness numbers, which makes their scores more ambitious and harder to audit. Apple’s is more transparent and more limited. If you have ever wondered why an Oura sleep score and an Apple Sleep Score disagree on the same night, this is why: they are not measuring the same thing.
We took that question apart in detail in how recovery scores actually work.
If your score is lower than you want
The weights make the priority order obvious, and we worked through it point by point in how to improve your Apple Watch Sleep Score. If the score is not appearing at all, start with Apple Watch not tracking sleep, which covers the four settings responsible for most cases.