Bottom Line First
  • The "new" study isn't new. It was published on 8 December 2025 in SLEEP Advances by a team at Oregon Health & Science University. The September 2026 headlines are a second wave of coverage
  • It compared counties, not people. Across 3,141 US counties, places where more adults reported sleeping under seven hours had shorter life expectancy โ€” a link seen in 47 of 50 states in 2025
  • "Better than diet" isn't in the data. The model had no measure of what anyone ate; its nearest variable was the county food-insecurity rate. Sleep ranked second to smoking in one model, and slipped behind obesity in the next
  • Individual-level evidence is real but modest. In a 2025 meta-analysis of 79 cohorts, under seven hours was linked to a 14% higher risk of death than 7โ€“8 hours. Low fitness shows far bigger gaps
  • What to do: aim for seven hours or more, keep a regular wake time, keep exercising, and if insomnia has lasted months, ask your doctor about CBT-I

"One Critical Factor Predicts Longevity Better Than Diet or Exercise," ScienceAlert announced on 22 September 2026. RealClearScience picked it up three days later as "Sleep Predicts Longevity Better Than Diet or Exercise," and the claim spread from there. It's a satisfying message: the thing you do lying down might beat the salad and the treadmill.

We found the paper. It's a real, peer-reviewed study, and its core finding deserves attention. But it compares US counties, not people, and the line about diet and exercise comes from the press release, not from anything the paper measured head to head. Below: what it did, what it can't show, what stronger evidence on sleep and death says, and what actually helps. Every figure is traced to its source.

47 of 50
US states where counties with more short sleepers had significantly shorter life expectancy in 2025 (McAuliffe et al., SLEEP Advances, December 2025)
Part one ยท The paper

The Study Behind the Headline, and Its Real Date

Evidence: Weak for individuals โ€” observational, county-level (ecological)

The paper is "Sleep insufficiency and life expectancy at the state-county level in the United States, 2019โ€“2025," by Kathryn McAuliffe, MSN, RN, of Oregon Health & Science University (OHSU) and seven colleagues, with sleep physiologist Andrew McHill as senior author. It was published in the journal SLEEP Advances on 8 December 2025 (McAuliffe, SLEEP Advances 2025). SciTechDaily covered it on 31 December 2025 under the headline "Sleep May Matter More Than Diet or Exercise for How Long You Live." The ScienceAlert story in September 2026 was a second round.

The sleep figures trace back to the CDC's Behavioral Risk Factor Surveillance System, a large annual telephone survey: the share of adults who report sleeping fewer than seven hours a night. The team lined that up against life expectancy in 3,141 counties, year by year, from 2019 to 2025.

First came simple correlations within each state. Counties with more short sleepers had shorter life expectancy; the share of states where that was significant ranged from 84% in 2019 to 100% in 2024, and in 2025 it held in all but three. Second, a "mixed-effects" regression across all the years that put insufficient sleep alongside other county-level factors: physical inactivity, food insecurity, health insurance, unemployment, high-school completion, social connections and smoking.

Figure 1 ยท McAuliffe et al., 3,141 counties, 2019โ€“2025

How strongly each county factor tracked life expectancy, in two models

MODEL 1 ยท EIGHT COUNTY FACTORS Smoking โˆ’0.31 Short sleep โˆ’0.17 MODEL 2 ยท + OBESITY AND DIABETES Smoking โˆ’0.28 Obesity โˆ’0.18 Short sleep โˆ’0.08

Regression coefficients (b) from the paper's text; all p < .0001. Longer bar = stronger negative link with county life expectancy. Other factors' estimates appear only in the paper's chart, so aren't shown.

In the first model, insufficient sleep had a coefficient of โˆ’0.17; only smoking, at โˆ’0.31, was larger. That's behind the press-release line that sleep "stood out more than diet, more than exercise, more than loneliness." In a second model the researchers added obesity and diabetes, which they treated as possible pathways between poor sleep and early death. Sleep stayed significant, but its coefficient fell by more than half, to โˆ’0.08, and obesity (โˆ’0.18) now ranked above it.

McHill was candid about being surprised. "I didn't expect it to be so strongly correlated to life expectancy," he said in OHSU's announcement, adding: "People really should strive to get seven to nine hours of sleep if at all possible." Sound advice. The question is what this study adds to it.

Part two ยท Reading it carefully

Why "Better Than Diet or Exercise" Doesn't Follow

Studies that compare places rather than people are called ecological studies. They're good for spotting patterns and steering public-health money, and weak at telling an individual what to do, because a county where many people sleep badly differs from its neighbours in dozens of ways at once. For this paper, that means five things.

1. There was no diet variable. The model's list of factors contains nothing about what people eat. The nearest item is food insecurity, which County Health Rankings โ€” the site the paper's data statement points to โ€” defines as the "percentage of population who lack adequate access to food." Its figure comes from Feeding America's Map the Meal Gap, estimated from Census and labour-market data. It tracks whether people can get enough food, not what they choose to eat. In earlier County Health Rankings releases it sat under a "diet and exercise" heading, which may be how "diet" reached the press release. Physical inactivity was in the model, but as the share of adults in a county reporting no leisure-time physical activity, from the same telephone survey โ€” not a measure of anyone's fitness.

2. The factors overlap heavily. Poverty, smoking, obesity, inactivity and short sleep tend to cluster in the same places. When you put many such overlapping numbers in one model, the individual coefficients become hard to read. A telling sign: the paper notes that some factors showed significant positive associations with life expectancy, including no insurance and unemployment. Nobody thinks losing your health insurance or your job helps you live longer. Odd signs like these are a classic symptom of overlapping factors in one model, and a reason not to read the ranking as a league table of causes.

3. The county sleep numbers are partly modelled. They aren't simple head counts from the survey. County Health Rankings explains that its county estimates come from the CDC's PLACES project, which uses a statistical method that "links BRFSS data with high spatial resolution population demographic and socioeconomic data" from the Census Bureau's American Community Survey. In other words, a county's sleep figure is shaped in part by its age mix and socioeconomic profile โ€” the same things that predict life expectancy.

4. The timing doesn't line up. A habit that shortens life should come before the deaths it causes. County Health Rankings' 2025 release, for example, pairs sleep data from 2022 with life expectancy calculated from deaths in 2020โ€“2022. If, as the paper's data statement suggests, its yearly figures came from these releases, then each year's comparison is mostly a snapshot of two things measured around the same time, not a cause followed by an effect.

5. Long sleepers counted as "sufficient." The measure lumps seven hours and eleven hours together as enough. The authors flag this themselves, noting that the categorisation of long sleepers as sufficient "may have impacted" the results, because long sleep is also linked to poorer health. They also acknowledge that they "cannot exclude the risk of potential unmeasured confounding," and that they lacked data on shift work, sleep disorders and psychiatric conditions.

Figure 2 ยท What was measured

The headline's words versus the study's variables

Headline saysWhat the study usedWhy it matters
Sleep% of adults in a county reporting <7 hours a night (age-adjusted), partly modelled from survey and Census dataSelf-reported; counts 9+ hours as "sufficient"; carries some of the county's socioeconomic profile
DietFood insecurity: % of people who "lack adequate access to food"A measure of access, estimated from economic data โ€” not of what people eat
Exercise% of adults reporting no leisure-time physical activityA county average from the same survey, not fitness
LifespanCounty life expectancyDescribes the population; deaths can overlap in time with the sleep survey
"Predicts better"Size of a coefficient in a model of overlapping county factorsRanking changed when obesity and diabetes were added

Sources: McAuliffe et al. 2025; County Health Rankings & Roadmaps measure definitions and 2025 data-source list (the paper's data statement points to County Health Rankings).

The headlineSleep predicts how long you'll live better than diet or exercise.
What the study foundUS counties with more short sleepers have shorter life expectancy. In one model, only smoking had a stronger link; in another, obesity did too. Diet wasn't measured.

None of this makes the finding wrong. A pattern that shows up in nearly every state, every year, is real, and a fair reason for public-health agencies to take sleep seriously. It just isn't evidence that, for you, an extra hour in bed will do more than eating better or getting fit.

Part three ยท Following people, not places

What Stronger Evidence Says About Sleep and Death

Evidence: Moderate โ€” meta-analyses of large cohort studies

For questions about individuals, the better tool is the cohort study: ask people how they sleep, then follow them for years. Dozens exist, and they've been pooled several times.

The classic analysis, by Francesco Cappuccio and colleagues, combined 16 studies with 1,382,999 participants followed for 4 to 25 years, during which 112,566 died. Short sleepers had a 12% higher risk of death than people sleeping a normal amount, and long sleepers a 30% higher risk (Cappuccio, Sleep 2010). A 2025 update in GeroScience pooled 79 cohort studies. Compared with sleeping 7โ€“8 hours, sleeping under seven was linked to a 14% higher risk of death, and nine hours or more to a 34% higher risk (Ungvari, GeroScience 2025).

Figure 3 ยท Individual-level evidence

Higher risk of death linked to short and long sleep, versus normal sleep

UNGVARI 2025 ยท 79 COHORT STUDIES Under 7 hours +14% 9+ hours +34% Reference: 7โ€“8 hours CAPPUCCIO 2010 ยท 1.38 MILLION PEOPLE Short sleep +12% Long sleep +30% Reference: normal sleep, as defined by each study

Pooled hazard ratios 1.14 and 1.34 (Ungvari) and relative risks 1.12 and 1.30 (Cappuccio). Observational: these show association, not proof that changing your sleep changes your risk.

Two things stand out. First, the risk curve is U-shaped: both too little and too much sleep are linked to earlier death, compared with the middle โ€” seven to eight hours in the newer analysis. Second, the long-sleep arm looks worse โ€” and that's the arm the county study counted as "sufficient." Observational data can't tell whether sleeping nine or ten hours is harmful in itself or whether people who are already unwell sleep longer. For most readers, the short-sleep figure โ€” a 12โ€“14% higher risk โ€” is the more useful number.

Hours may not even be the most important dimension. In 2024, a UK Biobank study measured sleep with wrist sensors for a week in 60,977 people rather than asking them, then followed them for an average of about six years. The most regular sleepers had a 20% to 48% lower risk of dying from any cause than the least regular, and regularity was a stronger predictor of mortality than duration (Windred, Sleep 2024). Our full guide to sleep and longevity covers this study and the duration curve in more depth.

The county pattern also has an obvious individual-level backdrop. In the CDC's 2014 survey of 444,306 adults, 65.2% reported sleeping at least seven hours. Healthy sleep was least common in the southeastern states and along the Appalachian Mountains, a distribution the CDC noted "is similar to geographic variations in prevalence estimates for obesity and diabetes and death rates from heart disease and stroke." Only 51.0% of people who were unable to work got enough sleep, against 64.9% of those employed (Liu, CDC MMWR 2016). Illness disrupts sleep, and county data can't separate cause from effect.

Part four ยท The comparison in the headline

Is Sleep Really Bigger Than Exercise?

Evidence: Different studies, different contrasts โ€” no clean ranking

No study we know of has ranked sleep, diet and exercise against each other in the same people. But you can set individual-level numbers side by side, with care.

One of the largest studies of measured fitness followed 122,007 adults who took a treadmill test at an academic medical centre between 1991 and 2014, with a median follow-up of 8.4 years. Compared with the fittest group, the least fit had a fivefold higher risk of death (hazard ratio 5.04). Even the modest step from above-average to below-average fitness carried a hazard ratio of 1.41 โ€” the same as smoking in that study (Mandsager, JAMA Network Open 2018). Set that against a 12โ€“14% higher risk for short sleep. Different studies, different comparisons, so this isn't a precise ranking either. But nothing in the individual-level studies we reviewed suggests sleep outranks being fit.

They're not rivals anyway. A meta-analysis of 66 studies found that regular exercise has small benefits for total sleep time and sleep efficiency, small-to-medium benefits for how fast people fall asleep, and moderate benefits for sleep quality (Kredlow, Journal of Behavioral Medicine 2015). Exercise is one of the ways to sleep better.

The honest headline: sleep belongs on the list with smoking, weight, diet and exercise, and it's easy to neglect. "Better than diet or exercise" goes beyond the evidence.

Part five ยท The useful part

What Actually Improves Sleep

If the study nudges you to take sleep more seriously, good. Here's where the evidence points.

Protect seven hours. The American Academy of Sleep Medicine and Sleep Research Society recommend that adults aged 18 to 60 sleep at least seven hours a night (as summarised by the CDC). That usually means setting a bedtime as deliberately as an alarm.

Fix your wake time. Given the UK Biobank finding on regularity, the simplest lever is getting up at the same time every day, weekends included.

Keep moving. Exercise improves sleep quality, and fitness has one of the strongest links to survival in the research.

If insomnia has lasted months, get the right treatment. The American College of Physicians recommends that all adults with chronic insomnia receive cognitive behavioural therapy for insomnia (CBT-I) as the first treatment, a strong recommendation based on moderate-quality evidence (Qaseem, Annals of Internal Medicine 2016). It comes ahead of sleeping pills. Talk to your doctor before starting or stopping any sleep medication.

Don't ignore snoring. Loud snoring, gasping at night or nodding off by day can signal sleep apnea. Raise it with your doctor.

For the full ranked list โ€” light, caffeine, alcohol, temperature and supplements โ€” see our sleep and longevity pillar. Waking in the small hours? Try our guide to waking at 3am.

Frequently Asked Questions

Does sleep matter more than diet and exercise for how long you live?
That isn't what the study showed. It compared US counties, not people, and its model had no measure of what anyone ate โ€” the nearest variable was the county's food-insecurity rate. Sleep had the second-largest coefficient after smoking in one model, but fell behind obesity once obesity and diabetes were added. Studies that follow individuals link low fitness to much larger differences in mortality than short sleep.
What did the Sleep Advances study actually find?
Across 3,141 US counties from 2019 to 2025, counties where more adults reported sleeping under seven hours had shorter life expectancy. The link was significant in 47 of 50 states in 2025. In a model with eight county-level factors, only smoking had a larger coefficient than insufficient sleep. It is an observational, county-level study, so it shows a correlation, not cause and effect.
Is it a new study?
No. It was published in SLEEP Advances on 8 December 2025 by researchers at Oregon Health & Science University and covered at the time. The September 2026 headlines are a second round of coverage of the same paper.
How much does short sleep raise the risk of dying?
In studies that follow individuals, modestly. A 2025 meta-analysis of 79 cohort studies linked sleeping under seven hours to a 14% higher risk of death compared with 7 to 8 hours, and sleeping nine hours or more to a 34% higher risk. These are associations; illness and other factors can affect both sleep and survival.
How many hours of sleep should adults aim for?
The American Academy of Sleep Medicine and the Sleep Research Society recommend at least seven hours a night for adults aged 18 to 60. A regular schedule matters too: in a UK Biobank study of 60,977 people, the most regular sleepers had a 20% to 48% lower risk of death than the least regular. If you have chronic insomnia, talk to your doctor about CBT-I, the recommended first treatment.

Our Sources

  1. McAuliffe KE, Wary MR, Pleas GV, Pugmire KES, Lysiak C, Dieckmann NF, Shafer BM, McHill AW. Sleep insufficiency and life expectancy at the state-county level in the United States, 2019โ€“2025. SLEEP Advances 2025;6(4):zpaf090. academic.oup.com
  2. OHSU News. Insufficient sleep associated with decreased life expectancy. 8 December 2025. news.ohsu.edu
  3. Neuroscience News. Insufficient sleep strongly predicts shorter life expectancy (OHSU release). neurosciencenews.com
  4. ScienceAlert. One critical factor predicts longevity better than diet or exercise, study reveals. 22 September 2026. sciencealert.com
  5. RealClearScience. Sleep predicts longevity better than diet or exercise. 25 September 2026. realclearscience.com
  6. SciTechDaily. Sleep may matter more than diet or exercise for how long you live. 31 December 2025. scitechdaily.com
  7. County Health Rankings & Roadmaps. Insufficient sleep: measure definition and methods. countyhealthrankings.org
  8. County Health Rankings & Roadmaps. 2025 measures, data sources and years of data. countyhealthrankings.org (PDF)
  9. County Health Rankings & Roadmaps. Food insecurity: measure definition and source. countyhealthrankings.org
  10. County Health Rankings & Roadmaps. Physical inactivity: measure definition. countyhealthrankings.org
  11. County Health Rankings & Roadmaps. Life expectancy: measure definition. countyhealthrankings.org
  12. Cappuccio FP, D'Elia L, Strazzullo P, Miller MA. Sleep duration and all-cause mortality: a systematic review and meta-analysis of prospective studies. Sleep 2010;33(5):585โ€“592. academic.oup.com
  13. Ungvari Z, Fekete M, Varga P, Fekete JT, et al. Imbalanced sleep increases mortality risk by 14โ€“34%: a meta-analysis. GeroScience 2025. link.springer.com
  14. Windred, Burns, Lane, Saxena, Rutter, Cain, Phillips. Sleep regularity is a stronger predictor of mortality risk than sleep duration: a prospective cohort study. Sleep 2024;47(1):zsad253. academic.oup.com
  15. Liu Y, Wheaton AG, Chapman DP, Cunningham TJ, Lu H, Croft JB. Prevalence of healthy sleep duration among adults โ€” United States, 2014. MMWR 2016;65(6). cdc.gov
  16. Mandsager K, et al. Association of cardiorespiratory fitness with long-term mortality among adults undergoing exercise treadmill testing. JAMA Network Open 2018. jamanetwork.com
  17. Kredlow MA, Capozzoli MC, Hearon BA, Calkins AW, Otto MW. The effects of physical activity on sleep: a meta-analytic review. Journal of Behavioral Medicine 2015. link.springer.com
  18. Qaseem A, Kansagara D, Forciea MA, et al. Management of chronic insomnia disorder in adults: a clinical practice guideline from the American College of Physicians. Annals of Internal Medicine 2016. acpjournals.org
The Bottom Line

Sleep is a pillar. Not a trump card.

The study behind the headlines found a real, consistent pattern: US counties where more people sleep under seven hours have shorter life expectancy. But it compared places, not people, never measured diet, and its ranking shifted once obesity and diabetes were added.

The evidence from individuals says short sleep is linked to a modestly higher risk of death, and that regular sleep may matter as much as long sleep. So get your seven hours, keep a steady wake time, and keep the salad and the walks too. They work better together.