Why Apple Health and your Apple Watch show different sleep numbers

You went to bed at 11pm. You woke up at 7am. Eight hours, right? Apple Watch says you slept seven hours and twenty-two minutes. Apple Health's sleep tab says seven hours and forty. Your third-party sleep app says six hours, fifty-five. None of them agree, and none of them are technically wrong — they're just counting different things.

The short version: “time in bed,” “time asleep,” and “Apple Watch sleep” are three different measurements. Most of the confusion comes from sources mixing them up.

What Apple Watch measures

Apple Watch uses the accelerometer (movement) and heart-rate sensor to estimate when you're asleep versus awake but still. It needs you to be wearing the watch and to have a Sleep Schedule set up in the Watch app. With a schedule, the watch will start recording when your scheduled bedtime arrives and stop when you get up.

Apple Watch tries to detect the moment you fell asleep (not just the moment you went to bed), and the moment you finally woke up (not the last time you stirred at 6:47 before dozing off again). The result is closer to “time asleep” than “time in bed.”

What Apple Health shows

Apple Health's Sleep view aggregates several inputs:

• Sleep schedule data from the iPhone (“you went to bed at 11pm, your bedtime alarm went off at 6:30am, you got up at 7”).
• Apple Watch detection as above, including its stage breakdown (REM, Core, Deep) on Series 8 and later.
• Third-party app contributions from any sleep app you've granted write access.

When the Sleep tab shows a number, it usually shows the longest contiguous “in-bed” period from the highest-confidence source, falling back to the iPhone schedule when the watch is missing.

Why third-party apps disagree

A third-party sleep app may use:

• Just the watch's data (matching the watch's number).
• Just the schedule data (matching the iPhone's in-bed range, which is wider than time asleep).
• A blend, with the app's own filter on top — e.g. excluding short awakenings, requiring a minimum movement threshold.

Two sleep apps reading the same HealthKit data can show different numbers because they apply different policies on top. That's normal, but it's worth understanding which policy each app uses.

What Nightpenny does

Nightpenny reads the standard HKCategoryTypeIdentifierSleepAnalysis records from HealthKit. For each night it sums the segments marked asleep (across REM, Core, and Deep when the watch differentiates them), and ignores the segments marked inBed but not asleep.

That gives you the “time actually asleep” number — usually shorter than the Apple Health Sleep headline and matching the Apple Watch's sleep duration on the watch face. If you switch to Nightpenny from an app that showed time-in-bed, your reported sleep will look 20-40 minutes shorter; that's the policy difference, not a regression.

Which number is “right”?

For tracking sleep debt — the running tally of how much restorative sleep you're getting versus your target — the “time asleep” number is right. Lying in bed reading at 11pm isn't sleep, even if your bedtime says it is.

For tracking sleep latency (how long it took you to fall asleep) or sleep efficiency (the ratio of asleep to in-bed), you want both numbers: time asleep divided by time in bed. The closer to 100%, the more consolidated your sleep.

For sleep stages, the watch is your only source. iPhone bedtime data alone can't see REM vs Core vs Deep — that needs the accelerometer + heart-rate signal.

A few practical tips

• If you don't wear the watch overnight, Apple Watch detection won't fire and the Sleep view falls back to your scheduled in-bed window. That number will consistently over-report your actual sleep.
• If the watch is on but the Sleep Schedule is off, the watch will still detect sleep based on movement and heart rate, but it won't know when you intended to be asleep. The detection still works; the “in bed” framing doesn't.
• Don't edit sleep data manually unless you have to. Manual edits override the sensor data and can't be undone cleanly. If the watch missed a night, log it via the app rather than rewriting the HealthKit record.
• Compare apps to themselves, not to each other. The trend matters more than the absolute number. If Nightpenny says your average has dropped from 7h 30m to 6h 45m over the last month, that's the signal — whether another app would call it 7h vs 6h 15m is policy noise.

FAQ

Why does Apple Health sometimes show more than the watch?

Because the Health Sleep view often shows the in-bed range from your schedule, not the watch's asleep estimate. The watch detects when you fell asleep; the schedule starts the window from when you intended to.

Can I trust the REM and Deep numbers from Apple Watch?

Roughly. The watch's stage detection has improved generation over generation but is still an estimate from movement and heart rate — not the gold-standard EEG a sleep lab would use. Treat the stage numbers as relative (more REM this week than last) rather than absolute.

Why does my watch sometimes miss the start of the night?

Usually low battery or charging interruptions. The watch needs at least ~10% to record overnight; charge it before bed or during dinner.

The full breakdown of which HealthKit signals Nightpenny reads, and how the math turns them into a 14-night balance, is on the methodology page.