Sleep Regularity, the Most Under-Dosed Sleep Input
Sleep regularity may predict health outcomes as strongly as duration. Learn to score your sleep timing from your tracker and retrain it in four weeks.

In this article
- 1.The Evidence That Sleep Regularity Rivals Duration
- 2.Why the Body Cares About Timing, Not Just Totals
- 3.What the Sleep Regularity Index Actually Measures
- 4.How the sleep regularity index works
- 5.Vendor consistency scores are not the SRI
- 6.Score Your Own Regularity From Wearable Exports
- 7.Pull 7 to 14 nights from your app
- 8.The three numbers to compute
- 9.Two worked examples
- 10.The Anchor Wake Time Protocol, Four Weeks With Tolerance Windows
- 11.Why wake time, not bedtime
- 12.The four-week progression
- 13.Edge Cases, Travel, Weekends, and Shift Work
- 14.When Rigidity Backfires, Orthosomnia and the Duration Override
- 15.The Seven-Day Starter Sequence and Re-Test
You have already bought the easy wins. Blackout curtains, a cool room, magnesium, the tracker that grades every night while you are unconscious. Meanwhile the cheapest variable in your entire sleep stack, the one that tracks with mortality risk in large accelerometer cohorts about as strongly as how long you sleep, drifts through your data unexamined. Friday bedtime is 1:30 a.m. Tuesday is 10:45 p.m. Sunday morning starts whenever it wants. Researchers call the stability of that timing sleep regularity, and for wearable owners it is the most under-dosed input left.
Stay in the loop.
Get the latest posts and exclusive content delivered to your inbox.
Join 50 readers. No spam. Unsubscribe in one click, anytime.
Four things follow that vendor posts skip. The actual evidence on whether regularity rivals duration, a ten-minute method to score your own sleep timing consistency from data you already own, a four-week anchor-wake block with explicit tolerance windows, and the trade-off nobody sells, which is when rigid consistency stops being optimization and starts costing you sleep.
The Evidence That Sleep Regularity Rivals Duration
The headline finding comes from the UK Biobank, where roughly 88,000 adults wore wrist accelerometers for continuous measurement. In a 2024 analysis of that cohort, people with more regular sleep-wake patterns showed lower all-cause and cardiovascular mortality risk over follow-up. The part that rearranges priorities is the comparison: regularity predicted mortality as strongly as sleep duration did, and in some models more strongly.
That inverts the hierarchy most sleep stacks are built on. Duration gets the supplement aisle, the evening wind-down, the sleep-maxxing threads. Regularity gets nothing, because there is no product to sell for going to bed at the same time.
More recent UK Biobank work extends the pattern from death to disease. UK Biobank multimorbidity analyses link irregular sleep timing to developing multiple cardiometabolic conditions, things like hypertension, diabetes, and heart disease, over follow-up, on top of the mortality findings.
Now the honesty clause. All of this is observational. Nobody randomized 88,000 people to regular or chaotic schedules, so confounding is possible, and people with very irregular lives may differ in ways the models miss. Two points soften that skepticism. The exposure was device-measured rather than self-reported, which removes the worst source of noise in sleep research. And the association appears across distinct outcomes, mortality, cardiometabolic disease, and academic performance in the original work, which is the kind of convergence that survives scrutiny better than a single flashy result.
Why the Body Cares About Timing, Not Just Totals
Correlation this consistent deserves a mechanism, and the mechanism is circadian alignment. Your body does not run glucose control, blood pressure regulation, or hormone release on total sleep time. It runs them on internal clock phase, and the clock shifts slowly.
Insulin sensitivity falls at internal night. Blood pressure follows a daily rise-and-dip rhythm. Cortisol and melatonin secretion are gated to phase, not to whether you happened to be asleep. When behavior lurches while the clock lags, you end up eating, digesting, and sleeping at the wrong internal times, repeatedly.
The lab version of this is ugly. In a controlled misalignment experiment where participants lived on shifted schedules, circadian misalignment impaired glucose regulation, raised blood pressure, and suppressed the satiety hormone leptin, among other effects, with total sleep held constant. An irregular week is a milder, repeated version of that same mismatch. Friday's 2 a.m. bedtime does not just shorten sleep, it asks a phase-gated body to process a pizza at internal 4 p.m. and a cortisol peak at the wrong hour.
What the Sleep Regularity Index Actually Measures
How the sleep regularity index works
The metric behind these cohort findings is the sleep regularity index, introduced by the original SRI paper from Phillips and colleagues in 2017. It compresses a week of movement data into one number, the probability that you are in the same state, asleep or awake, at any given clock time 24 hours apart. The scale runs from 0, meaning perfectly random timing, to 100, meaning perfectly regular.
For orientation, published cohort averages tend to land in the low-to-mid 60s. Scores in the 80s read as notably regular, and scores below the low 50s read as clearly irregular. Treat those bands as rough guides rather than clinical thresholds.
Vendor consistency scores are not the SRI
Your app probably shows something called consistency, and the sleep consistency score meaning depends entirely on the vendor. Oura's regularity explainer describes a two-week measure of bedtime and wake-time spread feeding a readiness contributor. Garmin and Whoop ship their own proprietary versions. These are directionally useful, they will drop when you rage against your schedule, but they are not interchangeable with the published index, which is computed from minute-level state data rather than two summary timestamps. That gap is exactly why the next section has you compute your own.
Score Your Own Regularity From Wearable Exports

Any device that timestamps bed and wake times holds enough data for a useful, research-adjacent score. You need 7 to 14 nights and about ten minutes.
Pull 7 to 14 nights from your app
- Apple Watch: export health data from the Health app, or use a third-party export app, and pull nightly bedtime and total sleep.
- Oura: generate a CSV from the website's export tool.
- Garmin: export sleep data from Garmin Connect.
- Whoop: use the in-app data export.
You only need two columns per night, when you fell asleep and how long you slept.
The three numbers to compute
- Sleep midpoint for each night equals bedtime plus half the total sleep time. A 23:30 bedtime with 7 hours asleep puts the midpoint at 03:00.
- DIY regularity percentage. Take the median midpoint across your nights, then count how many fell within plus or minus 30 minutes of it. Nine of 14 nights is 64 percent.
- Social jetlag, the classic mismatch between weekday and free-day timing. Take the absolute difference between your average weekday midpoint and your average free-day midpoint. Two hours means your weekend self lives two time zones from your weekday self.
- Day-over-day drift, the average absolute change in midpoint from one night to the next. This is the closest cousin to what the SRI rewards.
If you have wondered how to calculate the sleep regularity index at home, the honest answer is that the full version needs epoch-level data, as published SRI calculation methods describe. The midpoint version above is a simplification, but it moves when the SRI moves and it costs you a spreadsheet, not a grant.
Social jetlag deserves its own sentence because it is the component most weekend athletes carry. Cohort studies on social jetlag associate wider weekday-to-free-day gaps with worse cardiometabolic risk markers, and the associations are commonly discussed as the gap widens beyond about an hour.
Two worked examples
| Metric | Maya, weekend compressor | Ben, anchored |
|---|---|---|
| Weeknight pattern | 23:30 to 06:30 | 22:45 to 06:15 |
| Weekend pattern | 01:30 to 10:00 | 23:00 to 06:45 |
| Midpoint spread | 2 h 45 min | 30 min |
| Social jetlag | 2 h 45 min | ~20 min |
| Day-over-day drift | ~50 min | ~15 min |
| Within ±30 min of median | 8 of 14 nights (57%) | 12 of 14 nights (86%) |
Maya sleeps a healthy 7 to 8.5 hours every night. Her duration is fine, and she would pass any sleep-hygiene audit. Her timing tells another story, a nearly three-hour weekly migration that her tracker averages away. Ben has a boring schedule, which is the point. Rough heuristics for classifying yourself: drift under 30 minutes with social jetlag under an hour is tight, either measure in the 30-to-60-minute or one-to-two-hour range is moderate, and anything beyond that is loose.
The Anchor Wake Time Protocol, Four Weeks With Tolerance Windows

Why wake time, not bedtime
Bedtime is an output. It depends on sleep pressure, evening stress, and how the previous night went, which is why forcing it fails. Wake time is an input. You control it with an alarm, and it sets the timing of the strongest clock cue you own, morning light. Reviews of timed light identify bright light exposure as the primary lever for shifting circadian phase, and outdoor light, even on an overcast morning, typically delivers far more lux than a consumer sunrise lamp at pillow distance.
Hold the wake anchor and two things follow. Light lands at a consistent phase, and rising sleep pressure gradually pulls bedtime earlier on its own. You anchor the clock at the front and let the back end organize itself.
The four-week progression
| Week | Wake window | Bedtime rule | Supporting inputs |
|---|---|---|---|
| 1 | Anchor ±45 min | None. Sleep when sleepy. | 10 to 30 min outdoor light within an hour of waking; stop caffeine 8 to 10 hours before bed |
| 2 | Anchor ±30 min | Add a fixed wind-down cue | Same, plus dim household lights after sunset |
| 3 | Anchor ±20 min | Optional, shift wind-down 15 min earlier if duration is short | Same |
| 4 | Anchor ±15 min | Hold | Same |
Standing rules for the block:
- One anchor, seven days a week. The weekend is where schedules die.
- A bad night does not move the anchor. Wake at the window anyway, expect a rough day, and let the sleep pressure pay forward into an earlier, deeper night.
- Add sleep at the front, earlier bedtime or a short nap before 15:00, never a long morning lie-in.
- Expect weeks 1 and 2 to feel worse before better. Consolidating a drifting schedule briefly resembles mild sleep restriction.
This block is also most of the answer to how to fix social jetlag. Cap weekend lie-ins at an hour past the anchor and the weekday-to-weekend midpoint gap compresses on its own.
Edge Cases, Travel, Weekends, and Shift Work
Weekend recovery sleep. Sleeping in helps sleepiness, and it also re-shifts the clock, which is how Sunday insomnia and Monday grogginess get manufactured. If you are short on sleep, take it Friday evening in front, not Saturday morning in back, and keep Saturday and Sunday wake within an hour of the anchor.
A worked travel case. You fly two time zones east for three nights. Under 48 hours, do not convert at all, hold your wake anchor within an hour of home time and get outdoor light in the destination morning so the clock does not drift further. For longer trips, shift the anchor 30 to 60 minutes per day toward local time and move morning light with it. The error to avoid is anchoring fully to local time on day one, then flying home Thursday, which stacks two misalignments in one week.
Shift work. Consistency within your dominant shift beats forcing a daytime schedule your roster will not allow. If you rotate, forward rotation, afternoon into night into morning, costs less than backward. Protect the daytime sleep you actually get with blackout, a dark commute home, and light discipline at the wrong hours. An imperfect schedule held steady outperforms a perfect schedule abandoned every fourth week.
When Rigidity Backfires, Orthosomnia and the Duration Override
In 2017, a case series in the Journal of Clinical Sleep Medicine named orthosomnia, sleep perfectionism driven by tracker scores. It is the failure mode of everything above, and wearable owners are the population at risk.
Signals that consistency is now costing more than it pays:
- You check the sleep score during the night, or before getting out of bed, and it colors the morning.
- A 72 score produces anxiety that a 72 has no business producing.
- A late evening triggers abandoning the wind-down entirely, since the night is "already ruined."
- You decline social, family, or travel plans to defend the score.
The plateau principle matters here, too. The goal is a repeatable schedule, not a perfect one, and chasing perfection buys marginal regularity at the cost of stress that keeps you awake.
Two conditions override the protocol entirely:
- Active sleep debt. Illness, a heavy training block, a newborn, an acute deadline. When you genuinely need more total sleep, duration wins. Widen the window to ±60 minutes, protect totals, and re-tighten when the stressor clears.
- Tracking distress. When scoring itself becomes a sleep disruptor, cut measurement to the weekly re-check below. The data serves the schedule, never the reverse.
The Seven-Day Starter Sequence and Re-Test
- Days 1 and 2. Export your data and compute the three numbers. Classify yourself tight, moderate, or loose.
- Day 3. Pick an anchor within an hour of your current weekday wake time. Not aspirational, current. Set the window at ±45 minutes.
- Days 4 to 7. Outdoor light within an hour of waking, caffeine cutoff 8 to 10 hours before bed, no bedtime policing yet.
- Every Sunday. Recompute the three numbers in ten minutes. Weekly cadence, not nightly.
At the four-week re-audit, decide with data:
| If... | Then... |
|---|---|
| Drift under 30 min, duration held, mood steady | Tighten to the next window |
| Drift 30 to 60 min | Hold the current window another week |
| Duration dropped 30+ min per night, or dread of the alarm | Loosen one step, or widen to ±60 |
Maya from the worked example would land on the Friday-early-bedtime rule first, because her 2-hour-45-minute social jetlag is the drag, not her weeknight discipline. Ben holds. The reader from the opening paragraph, the one with the curtains and the drifting bedtime, now has a number where the vague guilt used to be, and a four-week block that moves it. Anchor the wake, let bedtime follow, measure weekly, and loosen the moment the schedule starts serving itself instead of your sleep.
Stay in the loop.
Get the latest posts and exclusive content delivered to your inbox.
Join 50 readers. No spam. Unsubscribe in one click, anytime.
About the author
Elena Park
Sleep and Recovery Specialist
Elena has spent a decade helping people fix their sleep and recover faster, pairing wearable and HRV data with the cold, heat, and breathwork protocols she tests on herself first. She writes recovery routines that hold up under real, busy lives.
Related Posts
Orthosomnia, When Your Sleep Tracker Creates Insomnia
Orthosomnia is insomnia your sleep tracker can create. Learn the mechanism, run a two-week blind test, and replace nightly scores with a quarterly audit.
Your Sleep Setup Cost per Night, Every Layer Ranked
Your sleep setup cost per night, ranked layer by layer: the mattress runs 22 cents, the Oura Ring 51 cents. See where the next dollar buys the most sleep.
Best Sleep Position by Symptom, With a Retraining Plan
Find your best sleep position for reflux, snoring, or shoulder pain, plus the mechanisms, graded claims, and a retraining plan that finally sticks.


