The Sleep Divorce Index, 2026

This is the single most useful thing on this page. Sleep divorce figures are not competing estimates of one number. They are answers to five genuinely different questions about behaviour, arranged here from strictest to loosest, plus one about attitude that is often quoted alongside them. Read them as a ladder, not a range.

• 1.4% - started and maintained for a year. A January 2023 SleepFoundation.org survey of 1,250 US adults found only 1.4% had begun sleeping separately and sustained it over the past year. The follow-on figures are quoted everywhere, and they deserve their denominators: 52.9% said sleep quality improved, by a reported 37 minutes a night, and 25.7% recoupled. Those are shares of the people who slept apart, not of US adults, and the maintainer cell is roughly 18 people self-reporting against a remembered baseline. A 37-minute figure carried to the minute off a base that size is false precision, and we would not have caught that if we had taken the number from an article instead of the survey.
• 18% - permanently, worldwide. ResMed's fifth annual Global Sleep Survey of 30,026 people across 13 markets found 18% of couples permanently sleep apart because of snoring and restlessness. Among them, 31% said the relationship improved and 30% said it worsened, which is close enough to a coin flip to be worth sitting with.
• 21% - currently maintain a separate bedroom. The International Housewares Association reported in January 2023 that one in five adults currently keep separate bedrooms from a partner, and that 46% of them attribute it to snoring or incompatible sleep habits. IHA does not publish its sample size, which is a real limitation.
• 31% - have ever done it. The AASM's 2025 Sleep Prioritization Survey of 2,007 US adults, fielded 5-13 June 2025 by Atomik Research with a margin of error of plus or minus 2 points, found 31% have opted for a sleep divorce at some point.
• 35% - the 2023 figure, and a different question. The AASM's 2023 survey of 2,005 adults asked about sleeping in another room and split the answer into "on occasion" (20%) and "consistently" (15%). Adding those to 35% is what produced the "more than a third of Americans" headline that still circulates. It is a valid number for the question asked. It is not the same question as 2025's.
• 48% - would like it normalised, which is not behaviour. A January 2024 Wakefield Research survey of 2,000 US adults found 48% agree we should normalise separate bedrooms for married couples. That is an attitude, and attitudes are consistently more permissive than behaviour. It belongs on a different ladder entirely.

The practical consequence: if you see "sleep divorce is up" or "one in three couples," ask which rung of this ladder the number came from. In most coverage, two different rungs are quoted in the same paragraph.

Who actually sleeps apart, the young or the old? The published data says both, and the contradiction is the finding rather than a flaw in it.

• AASM 2025 says younger. Adults aged 35-44 were the most likely to report a sleep divorce at 39%, and adults 65+ the least likely at 18% (AASM 2025). The 2023 wave agreed in direction: millennials 43%, Gen X 33%, Gen Z 28%, boomers 22%.
• IHA says older, by a factor of three. Adults 65+ were three times as likely to maintain separate bedrooms (29%) as adults 18-34 (10%), and 75% of those keeping separate bedrooms had lived together for a decade or more (IHA 2023).
• The oldest survey agrees with IHA. The Better Sleep Council's September 2012 survey of 542 partnered US adults found 16% of those 55+ slept in separate bedrooms against 3% of 18-34s (Better Sleep Council). That survey is now well over a decade old and we cite it for direction only, not for its level.
• The resolution is architectural, not generational. Occasional accommodation needs nothing but a couch, and younger couples report plenty of it. Maintaining a separate bedroom needs a spare bedroom. Housing tenure, not attitude, may be doing most of the work in the older cohorts. Nobody has tested this directly, which is itself worth noting.

The AASM has published a sleep divorce figure three years running, and that looks like a time series. We went into this report intending to chart it. Then we read the three survey instruments and found we could not, honestly. Here is the record as three separate readings, with the wording change disclosed.

Note the direction of the problem. The 2024 and 2025 wording is broader than 2023's, since it counts another bed in the same room, yet it returns a lower number. That is the signature of an instrument change, not of behaviour. All three waves were run by Atomik Research with a margin of error of plus or minus 2 points, so sampling noise does not explain a six-point move either.

We went looking for the sources behind the most-repeated claims in this topic. These five did not survive the walk back. We are naming them because the recycling is the problem: each one is now quoted with total confidence by outlets that never opened the original.

• "Snoring causes divorce." We could not find a primary study behind this, and we looked hard. The citation trail runs to a 2006 narrative review (Shah, Roux and Mohsenin, Treat Respir Med), which states in passing that patients "often have marital problems with increased divorce rates" and presents no data of its own for that claim. Being straight about the limits of our own check: that paper's full text and reference list sit behind a paywall we did not clear, so we cannot tell you what it cites in turn. What we can tell you is that no study we found has followed couples to separation, and that a claim this confident should not be this hard to source.
• "A snorer costs their partner an hour of sleep a night." The 62-minute figure was never measured, and the authors said so. In the 1999 Mayo Clinic study, ten couples were recorded for a single night, CPAP was applied halfway through, and spouses' median sleep efficiency rose from 74% to 87%. The paper's own abstract then states: "Assuming that 480 minutes were spent in bed for sleep, a 13% improvement in sleep efficiency ... translates to an additional 62 minutes of sleep per night." The word "assuming" is right there, in the most-read line of the paper. The distortion is not Beninati's, it is ours. It happened downstream, as each citer dropped the assumption and the conditional until a projection from ten couples over half a night became a fact about your marriage. Worth adding: those are medians, and multiplying a median by 480 is not a valid operation.
• "Almost half (48%) of 35-44-year-olds" (AASM 2025). This one is our own find, and it is the tidiest example on the list, because the error and its correction sit on the same page. AASM's 2025 report says "almost half (48%)" of 35-44-year-olds sleep apart, in a callout box. Its own age table puts that group at 39%. The 48% is what you get by adding the two answer options together (18% another bed in the same bedroom, plus 30% another space). The 39% is AASM's own "either/or" row, which nets out the people who picked both. The higher number is in the sentence built to be quoted; the lower one is in the table. We use 39% throughout this page.
• "23% of married couples do not sleep side by side, up from 12% in 2001." This one is a homebuilder citing itself. It appears in a 2011 trade-magazine article attributed to "research cited by Del Webb," a company that was at the time selling dual-owner-suite homes. It is frequently relaundered as a finding of the National Association of Home Builders. We could not find any NAHB source for it, or any primary source at all.
• "15% of Brits sleep apart." YouGov, surveying Great Britain in 2018, asked something else entirely: what people would prefer if cost and space were no object and there were no impact on the relationship. 15% said they would rather sleep in a different bed or room. 76% said they would still share. That is a hypothetical preference, and it is routinely reported as behaviour.

None of this means sleeping apart is rare or that snoring is harmless. It means the evidence for both is thinner and more careful than the coverage suggests, and that a lot of confident numbers have no floor under them.

Strip out the surveys and the press releases and ask what has been measured in a lab or a clinic. The honest answer is narrower than you would expect, and the direction of the best evidence is not the direction of the popular story.

• Partners of snorers report worse sleep and more daytime symptoms. The largest dataset here is Ulfberg and colleagues, a cross-sectional questionnaire of 1,032 Swedish women aged 30-64. Women living with heavy snorers more often reported insomnia symptoms, morning headache, daytime sleepiness and fatigue, apparently in a dose-response pattern, whether or not they snored themselves. It is correlational and both exposure and outcome are self-reported.
• In that same study, separate bedrooms did not help the women who tried them. Ulfberg's paper reports that sleeping in separate bedrooms "did not seem to give the women any alleviation." That is the largest partner-burden dataset in this literature, and it cuts against the premise of the entire sleep divorce conversation. We flag it rather than bury it. It is one study, self-reported, and not a randomized comparison.
• Partners report feeling better after treatment. Objective sleep is another matter. Two before-and-after studies in Chest, Parish and Lyng (54 couples) and Doherty and colleagues (55 couples), both found bed partners' sleepiness scores and quality of life improved once the snorer started CPAP. Neither had a placebo arm. The one trial that did, McArdle's randomized crossover in Thorax, found partners' subjective sleep improved but their objectively recorded sleep did not differ from placebo.
• Marital quality and treatment adherence are tangled, and the bigger study is the less flattering one. A small prospective study (Baron et al., 22 followed up) found marital conflict predicted CPAP use and fell after three months of treatment. A larger analysis of 194 participants from the APPLES trial (Batool-Anwar et al.) found no significant difference in marital quality between CPAP-adherent and non-adherent patients.
• Sleeping together has a measurable upside, in couples without a snorer. A sleep-lab study by Drews and colleagues found co-sleeping was associated with around 10% more REM sleep and less fragmented REM, alongside more limb movements. The couples were young and healthy. This is not evidence that bed-sharing survives a snorer, and it should not be used that way.

Snoring can be a symptom of obstructive sleep apnea, which is a medical condition with cardiovascular consequences and is not fixed by moving to the spare room. If you or your partner snore loudly, stop breathing during sleep, or are sleepy during the day, talk to a doctor and ask about a sleep study. A separate bedroom addresses the noise. It does not address the airway.

The strongest evidence that sleep divorce is not a new behaviour is that we already did it, at scale, and thought it was progress. Hilary Hinds' Wellcome-funded A Cultural History of Twin Beds (Bloomsbury) traces the arc: for the better part of a century, forward-thinking couples regarded sharing a bed as old-fashioned and unhealthy. The dated milestones below are drawn from the Guardian's account of the book, which quotes it directly. We have not put a precise start and end year on the era, because the sources we could open do not agree on one.

• 1892: the Yorkshire Herald proclaims that "the twin-bed seems to have come to stay, and will no doubt in time succeed the double bed in all rooms occupied by two persons." It was a prediction, and it did not come true. But it tells you which way the arrow was thought to be pointing.
• 1919: Dr Edwin Bowers, in Sleeping for Health, writes that "separate beds for every sleeper are as necessary as are separate dishes for every eater." They "promote comfort, cleanliness, and the natural delicacy that exists among human beings."
• 1920s: twin beds are seen as a fashionable, modern choice. This is the peak of the inversion. Sharing was the dated option.
• 1930s: twin beds are commonplace in middle-class households. Not universal, and never a majority of bedrooms, but unremarkable.
• 1940s: the turn. Twin beds now "occasion an unmistakable curl of the lip" and are "no longer the preserve of the health-conscious forward-thinking middle classes." Note what the sneer is about: this is class, not sex. The reading of separate beds as a verdict on the relationship comes later still.

The behaviour did not change nearly as much as its meaning did. What we now call a sleep divorce, and treat as a confession, our great-grandparents called a twin set and treated as an upgrade.

This is the thinnest part of the evidence base, and we would rather say so than fill it. Two findings are solid. Most of what circulates is not.

• Global, 18%. ResMed's 2025 survey of 30,026 people across 13 markets found 18% of couples permanently sleep apart (ResMed). ResMed makes CPAP machines, which is worth knowing when reading its sleep research. The 2026 wave dropped the sleep divorce question, so this is the current figure.
• Taiwan, and a result that complicates the wellness framing. A 2025 study in BMC Public Health found 59.9% of older Taiwanese couples shared a bed, and that sleeping in separate rooms was associated with lower psychological well-being (beta = -0.12, p < .01). Relationship quality measured a decade earlier enters only in the paper's sensitivity analyses, not the main model. It is cross-sectional, so it cannot settle which way the causation runs, and unhappy couples moving apart would produce the same result. It is still the clearest counterweight to the idea that sleeping apart is straightforwardly good for you.
• Japan: we could not verify any of it. The widely repeated claim that Japanese couples sleep separately at unusually high rates traces, in every instance we followed, to blogs, social posts and comment threads. We found no citable Japanese survey. We are reporting the gap rather than passing along the claim.
• UK, Germany, Scandinavia: no usable behavioural data found. The only UK number in circulation is the YouGov preference item covered above, which does not measure what people do.

The most striking thing about assembling this index was how much of the conversation rests on questions no one has actually asked. These are open, and we would like them closed.

• Who initiates. It is asked constantly in coverage and comment sections. We found no survey that measures it. Not AASM, not ResMed, not IHA, not SleepFoundation.org.
• What share of sleep divorces are snoring-driven versus schedule-driven. IHA's 46% for "snoring or other incompatible sleep habits" bundles two very different problems into one answer, and it is the closest anyone gets.
• Whether the snorer in a sleep divorce was ever screened for apnea. Every organisation publishing sleep divorce data pivots to apnea awareness in its commentary. None of them report what proportion of these couples ever had the snorer evaluated.
• Whether it lasts. Only SleepFoundation.org has published a durability measure, and its answer, 1.4% maintained over a year with 25.7% recoupling, is the single most surprising number in this whole topic. One survey is not enough to hang that on.