How Accurate Are Sleep Trackers?

Total sleep time: reasonable. Sleep stages: the least reliable thing they show you, and the number people fixate on.

What this is

Consumer sleep trackers are validated against polysomnography — the laboratory reference with brain, eye and muscle recording. A multicentre validation of eleven wearable, nearable and "airable" trackers analysed hundreds of thousands of epochs and found wide performance variation in sleep-stage classification, with the best device scoring well short of reference agreement and the worst substantially worse.

Total sleep duration is the most reliable output, typically within a modest error against polysomnography. Stage breakdowns — the deep-sleep and REM figures that dominate the app screen — are the weakest, with devices systematically over- and under-estimating in different directions.

What the evidence says

  • Total sleep time is reasonably accurate

    Gold· human data

    The most dependable metric across validation studies, usually within a modest margin of polysomnography. It's also the metric that matters most.

  • Sleep stage estimates are unreliable

    Gold· human data

    Multicentre validation found wide variation between devices and substantial deviations for stages, with light sleep overestimated and deep sleep underestimated in some devices by large margins.

  • Trends over weeks are more useful than nightly numbers

    Silver· human data

    If a device's error is consistent, the trend carries information the single night doesn't. Sensible use of a noisy instrument.

  • Tracker feedback can worsen sleep anxiety

    Silver· human data

    Clinicians describe orthosomnia in practice, with tracker data reinforcing rumination and unhelpful sleep behaviours. Case-based rather than trial evidence, and worth taking seriously.

  • You can optimise your deep sleep percentage

    Not supported

    You'd need an accurate measurement first. Chasing a number the device can't measure well is optimisation theatre.

What this means in practice

  • Use total sleep time and consistency of timing. Ignore the stage pie chart.
  • Compare weekly averages, not individual nights.
  • If the score contradicts how you feel, trust how you feel.
  • Take the device off for a fortnight if bedtime has become a performance.
  • For anything clinical — loud snoring, breathing pauses, chronic insomnia — see a GP. No tracker substitutes for assessment.

When to see a doctor

  • Snoring with pauses or gasping — get assessed regardless of what a tracker says.
  • Insomnia lasting over three months.
  • Anxiety about sleep that has grown since you started tracking.

The honest summary

Trust the duration, ignore the stages, watch the weekly trend, and take the thing off if it's making you anxious. The most over-interpreted number in consumer health is the deep-sleep figure.

Work it out

See also

Frequently asked questions

Can a smartwatch measure deep sleep accurately?
Not well. Multicentre validation against polysomnography found sleep-stage classification varied widely and fell well short of reference agreement, with some devices overestimating light sleep and underestimating deep sleep by large margins.
Which sleep metric should I actually use?
Total sleep time and the consistency of your sleep and wake times. Both are reasonably measured and both are what the health evidence is built on.
Can my ring detect sleep apnoea?
Some devices flag breathing disturbances as a screening prompt, which can be genuinely useful for getting people to a doctor. Diagnosis requires a proper sleep study.

Sources

Written and reviewed by Mathew Beale, MSc Biotechnology, University of Reading.

Last reviewed: