In one sentence
The probability, scaled 0 to 100, that you were in the same state, asleep or awake, at any two moments 24 hours apart over your last 28 nights: the same schedule every day scores 100, and sleep scattered at random scores near 0.
Why it matters
The index was introduced in a 30-day study of 61 students: in the least regular fifth, the body's melatonin onset came two and a half hours later than in the most regular fifth (00:08 against 21:32), the daily rhythm of sleepiness peaked later, and the light they lived in had a flatter rhythm (102 against 179 lux of amplitude); a model of the circadian clock attributed the delay mainly to their light patterns, and regularity correlated with academic performance (Phillips 2017). The reason it carries the most weight in VT's sleep score came later. In 60,977 UK Biobank adults with more than ten million hours of wrist-accelerometer recording (median SRI 81, interquartile range 73.8 to 86.3) and 1,859 deaths over a mean of 6.3 years, the four more regular fifths had 20 to 48 percent lower all-cause mortality than the least regular fifth, with the most regular fifth 30 percent lower after full adjustment; cancer mortality was 16 to 39 percent lower and cardiometabolic mortality 22 to 57 percent lower; and regularity was a stronger predictor of mortality than sleep duration when equivalent models were compared, with duration adding nothing to the regularity model (Windred 2024). The Methods article on the Science hub sets out the caveats, chief among them that the regularity-beats-duration claim rests on a model comparison and a test that did not reach significance.
What moves it
The clock, not the hours: weekend lie-ins, late nights, shift work, travel across time zones, a newborn, and naps, which count as sleep at an hour you are usually awake. It does not reward long sleep; six regular hours score higher than eight irregular ones, and the two numbers are scored separately in the Sleep Health Index for that reason. It moves slowly because it is 28 nights, so one late weekend costs a point or two and a month of them costs a band.
How to measure it well
Wear the same device every night; the index is computed from when you fell asleep, when you woke, and the awakenings between, at one-minute resolution, so a device that sits in the charger is a missing night, which is skipped rather than counted as irregular. The original study used sleep diaries; VT computes the index from your device's nights, and Night Log taps are shown on the night's chart but do not change it. A one-off device swap is set aside for that night, as every sleep reading is.
How Vitality Tracker uses it
Regularity is 35 percent of the Sleep Health Index, the heaviest component, and the index's regularity multiplier is built from it: a soft multiplier that reaches full strength only when low regularity persists across two consecutive 28-night windows, so that a disrupted month is treated as an episode, not a verdict. The Sleep page shows your index over the last 14 nights; the Sleep Health Index scores it over 28. The raw index is scored on the UK Biobank distribution, as published: an SRI of 71.65, the boundary of the least regular fifth, earns 50; 78.56 earns 70; 83.16 earns 85; 87.31, the boundary of the most regular fifth, earns 95; 92 and above earns 100, and 40 and below earns 0. The circadian-alignment component's second tier, used when no chronotype questionnaire has been completed, reads the same nights differently, each night's mid-sleep against your own usual; and the nightly sleep score's schedule-consistency component is the single-night version of the regularity idea. Shown with where you sit in the UK Biobank fifths.
How far to trust the number
Calculated from device timing, which is the part of a sleep record consumer devices get most nearly right: the index depends on when the big transitions happen, not on how the night was staged, and a device that misses the same short awakenings every night changes it little. Two caveats. The reference fifths come from one week of recording in adults averaging 63 years old, and VT computes over 28 nights, so your figure is comparable with the reference in kind, not exactly. And the index is a probability over pairs of days, so a single unusual night moves it less than you might expect and a changed schedule takes two to four weeks to show in full.
Sources
- Phillips AJK, et al. Irregular sleep/wake patterns are associated with poorer academic performance and delayed circadian and sleep/wake timing. Sci Rep 2017;7:3216. doi:10.1038/s41598-017-03171-4
- Windred DP, et al. Sleep regularity is a stronger predictor of mortality risk than sleep duration: a prospective cohort study. Sleep 2024;47:zsad253. doi:10.1093/sleep/zsad253
- Vitality Tracker Sleep Scoring Methodology, open specification v2.0.1, public edition. doi:10.5281/zenodo.23223863.