A Harvard study published in January 2026 followed 111,000 people for over 30 years. The conclusion: it's not just how much physical activity you do that matters. It's also how varied it is. People who practice the widest range of activities have a 19% lower risk of premature death, independent of how much total time they spend exercising.
It's not how much you move. It's how.
For years, the standard advice has been "30 minutes of exercise a day." That's right, but incomplete. A new American study adds a dimension that changes the picture: variety.
Researchers from the Harvard T.H. Chan School of Public Health analyzed the habits of 111,467 adults (70,725 women and 40,742 men) followed for over three decades. All were healthy at baseline: no cancer, no diabetes, no known cardiovascular or respiratory disease. Every two years, they reported their physical activities and the time spent doing them.
In total: more than 2.4 million person-years of follow-up, and 38,847 deaths recorded. A massive sample that allows for solid conclusions.
The headline result
People who practiced the greatest variety of activities had a 19% lower risk of premature death compared to those with the least variety. And this held even after accounting for total time spent moving.
In other words: at equal exercise volume, someone who walks, cycles, gardens, and does some strength training does better than someone who only runs, even if they run a lot.
Which activities stand out?
The study compared the most active practitioners to the least active for each activity type. The percentages below show the reduction in mortality risk for the most active:
Walking : 17% lower risk (the strongest effect, and the most accessible activity)
Tennis, squash, racquetball : 15% lower
Rowing, calisthenics : 14% lower
Weight/resistance training : 13% lower
Running : 13% lower
Jogging : 11% lower
Climbing stairs : 10% lower
Cycling : 4% lower
One activity in the study showed no statistically clear benefit on all-cause mortality: swimming. The researchers don't offer a definitive explanation, they simply note that the effect, if any, is too small to be detected with confidence.
The counter-intuitive detail: a ceiling
The benefits of physical activity don't keep rising indefinitely. For most activities, the protective effect rises quickly with the first hours per week, then levels off. Beyond a certain threshold, doing more doesn't add much.
For walking, for example, most of the benefit appears with the first regular kilometers, you don't have to be a marathoner.
🔍 Key takeaways
→Moving is good. Moving in varied ways is better: 19% lower mortality among the most diversified.
→Walking is the most powerful activity (–17%) and the simplest to integrate.
→Beyond a certain threshold, doing more provides little extra benefit, better to switch activities than double the dose.
→The study is observational: it shows an association, not definitive proof of cause and effect.
✦ Free newsletter
Want the next article straight to your inbox?
Free · No spam · Every two weeks.
✦ Intermediate summary
Han et al. published in BMJ Medicine (January 2026) a prospective study combining two reference cohorts, the Nurses' Health Study and the Health Professionals Follow-Up Study, covering 111,467 adults followed for over 30 years. Beyond total volume of physical activity, it is the diversity of activities practiced that proves predictive of lower mortality. Subjects in the most diversified quintile have a 19% lower risk of all-cause death after adjusting for total activity volume. Most activities studied show non-linear dose-response relationships, with a ceiling effect.
Two cohorts, thirty years of follow-up
The study draws on two massive prospective cohorts, among the oldest and best-documented in nutritional and behavioral epidemiology:
Nurses' Health Study (NHS), 70,725 American nurses followed from 1986 to 2018.
Health Professionals Follow-Up Study (HPFS), 40,742 male health professionals followed from 1986 to 2020.
All participants were free of cancer, diabetes, cardiovascular, respiratory and neurological diseases at baseline. They reported their physical activity habits every two years: type, duration, frequency. The NHS captured up to 11 distinct activities, the HPFS up to 13.
The unit used is the MET-hour per week. The MET (Metabolic Equivalent of Task) compares an activity's energy expenditure to that at rest. Brisk walking is around 3.5 to 4.5 METs, about four times resting expenditure. The method standardizes very different activities on a common energy scale.
Quantity vs variety: two distinct signals
The researchers first measured the association between total amount of activity and mortality. As expected, the relationship is clear: more movement, less premature death. But the hazard ratio (HR) doesn't decrease linearly, it drops quickly at the first activity steps, then plateaus. Beyond a certain threshold, the additional effect becomes marginal.
The most original finding concerns variety. The researchers built a simple score: how many different activities a person regularly engages in. Then they divided the participants into five groups based on this score.
Subjects in the most diversified quintile showed a HR of 0.81 for all-cause mortality (19% lower risk) compared to the least diversified, after adjusting for total activity volume. Variety is therefore an independent predictor. For cause-specific mortality: 13–41% lower risk depending on cause (cardiovascular, cancer, respiratory, other).
Activity by activity
Adjusted hazard ratios comparing the most active group to the least active for each activity:
Cycling: HR 0.96 (0.93–0.99), statistically significant but modest.
Swimming: HR 1.01 (0.97–1.05), no significant association detected.
The swimming result is notable. Several hypotheses exist in the literature, but this study doesn't settle the question. The authors note that swimming was less commonly practiced in the cohorts studied, reducing statistical power. The cautious conclusion: we cannot claim a strong protective effect on mortality, but neither can we rule it out.
Why variety?
Anna Whittaker, professor of behavioral medicine at the University of Stirling (not involved in the study), commented to the Science Media Centre: "This is likely due to the different types of activity having different physiological effects and helping to meet all of the aspects currently outlined in physical activity guidelines (i.e. moderate intensity exercise, resistance exercise, vigorous intensity exercise, flexibility work, recovery activities)."
In other words: walking maintains the cardiovascular system but doesn't build muscle. Strength training preserves bone density but doesn't tax the aerobic system. Tennis adds coordination and explosiveness. Stacking activity types ticks more physiological boxes.
🔍 Key points
→2 cohorts × 30 years × 2.4 million person-years = very solid statistical foundation.
→Variety and quantity are two independent predictors of mortality.
→Dose-response relationships are non-linear : likely ceiling effect around the recommended doses (7.5–15 MET-h/week).
→Observational study, self-reported activity, likely biases, but consistent with broad literature.
✦ Bottom line
The classic guidelines tell you how much. This study adds the what.
International guidelines (WHO, CDC) have long recommended 150–300 minutes of moderate activity per week. This study doesn't overturn those recommendations, it adds a strategic layer. At constant total time, it's better to spread across multiple types of activity than to concentrate on a single one. For most people, that means combining daily walking, some strength training, and a more dynamic activity (tennis, running, or stair climbing). Less a revolution than a refinement, but an actionable, free, no-equipment refinement.
Han H, Hu J, Lee DH et al. (BMJ Medicine, 2026; DOI: 10.1136/bmjmed-2025-001513) examine the associations between long-term engagement in individual physical activities and activity variety with mortality, using the Nurses' Health Study (1986–2018, n = 70,725 women) and Health Professionals Follow-Up Study (1986–2020, n = 40,742 men). Across 2,431,318 person-years of follow-up, 38,847 deaths occurred (9,901 cardiovascular, 10,719 cancer, 3,159 respiratory). Adjusted hazard ratios show inverse, non-linear associations between individual activities and all-cause mortality (except swimming: HR 1.01; 95% CI 0.97–1.05). A simple variety score, independent of total volume, predicts a 19% lower mortality (most vs least diversified quintile). An editorial correction was published 9 April 2026 (BMJ Med 5(1):e001513corr1).
Methods: stratified Cox models and variety score
The authors use Cox proportional hazards models with age as the time scale, stratified by cohort and calendar year. Physical activity is expressed in MET-hours/week, calculated as the product of self-reported duration and the MET value assigned to each activity. Variables are reassessed every 2 years, allowing for time-updated exposure tracking.
The variety score is a simple count of the number of distinct activities practiced above a minimum engagement threshold. Participants are then grouped into quintiles. The main analysis adjusts for: age, BMI, smoking, alcohol, diet quality (AHEI), family history, menopausal status (NHS), hormone therapy, and, critically, total activity volume. This adjustment is what isolates the variety-specific effect.
Total activity levels and variety scores are categorized into quintiles. For walking and stair climbing: quartiles. For other activities, due to heavy zero-skewing, non-practitioners serve as reference and practitioners are split into tertiles.
Detailed results
Pooled hazard ratios (NHS+HPFS) for all-cause mortality, highest vs lowest category:
Swimming: HR 1.01 (0.97–1.05), no significant relationship.
Variety, after full adjustment including total volume: HR 0.81 (≈19% lower risk) for all-cause mortality. Cause-specific: 13–41% lower risk depending on cause (cardiovascular, cancer, respiratory, other).
Modeled dose-response curves (restricted cubic splines) are non-linear for nearly all activities, with an inflection point around 7.5–15 MET-h/week, consistent with prior literature (Arem et al., JAMA Intern Med, 2015) suggesting a plateau beyond WHO recommendations.
Strengths and limitations
Strengths. Sample size (111,467 participants), follow-up duration (~30 years), repeated exposure assessment every 2 years (limiting misclassification), extensive adjustment for known behavioral and clinical confounders, two independent populations (women/men) with convergent results.
Main limitations.
Self-reported activity, not measured by accelerometer. As noted by Tom Yates (University of Leicester) at the Science Media Centre, this tends to underestimate true associations. Studies using objective measures of total activity suggest all-cause mortality risk is halved between active and inactive populations, far more than the paper's figures suggest.
Homogeneous population: American health professionals, predominantly white, high socio-economic status. Generalization to other populations needs validation, though physical activity–mortality associations are remarkably consistent across cohorts (cf. Patel et al., JAMA Netw Open, 2022 on 272,550 older adults).
Residual confounding: despite adjustment, practitioners of diverse activities may differ on unmeasured dimensions (disposable income, facility access, intrinsic motivation, social support). Variety could be a marker of broader well-being rather than a direct causal factor.
Reverse causality: early deaths in the cohorts may follow activity reduction due to subclinical conditions. The authors performed sensitivity analyses excluding the first years of follow-up without major changes to their estimates.
Special case for swimming: low prevalence (12% of participants) limits statistical power. Should not be interpreted as absence of benefit.
Technical note. An editorial correction (BMJ Med 5(1):e001513corr1, 9 April 2026) has been published, the corrected version should be consulted to cite final estimates.
🔬 Public health relevance
→Current guidelines (WHO 2020) primarily frame quantity (150–300 min/week moderate or 75–150 min/week vigorous). Adding a diversity objective to guidelines could be considered.
→Clinical implication: advise patients already active but focused on a single modality to broaden their palette before increasing volume.
→The dose-response plateau challenges the "always more" narrative and supports a satisficing approach to physical activity.
→Confirmation expected via cohorts using objective measurements (UK Biobank with accelerometers, NHANES).
✦ Bottom line
An open question for the field: is variety causal or merely a marker?
Han et al. provide a robust signal but don't definitively settle the mechanistic question. Does activity diversity capture more complete physiological coverage (cardiovascular + metabolic + musculoskeletal + neurocognitive), making variety causal? Or is variety a proxy for broader healthy behaviors (curiosity, social opportunities, economic resources) that co-determine longevity, in which case artificially increasing variety without changing those determinants would have little effect? The answer has different clinical and policy implications. Randomized trials over 30 years are unrealistic, but causal analyses (Mendelian randomization on proxies for exercise variety) could clarify the mechanism. For now, the practical translation is safe: at constant time, varying costs nothing and has no demonstrated risk.
Primary source
Han H, Hu J, Lee DH, Zhang Y, Giovannucci E, Stampfer MJ, Hu FB, Hu Y, Sun Q. Physical activity types, variety, and mortality: results from two prospective cohort studies. BMJ Medicine. 2026 Jan 20;5(1):e001513. doi:10.1136/bmjmed-2025-001513. Open access · CC BY-NC 4.0. Correction published 9 April 2026.
Key references
[1]Harvard T.H. Chan School of Public Health press release. Exercise variety,not just amount,linked to lower risk of premature mortality. 21 January 2026. hsph.harvard.edu ↗
[2]BMJ Group press release. Mix of different types of physical activity may be best for longer life. 21 January 2026. bmjgroup.com ↗, With detailed activity-by-activity figures.
[3]Expert reactions (Prof. Anna Whittaker, Prof. Tom Yates). Science Media Centre, 20 January 2026. sciencemediacentre.org ↗
[4]Patel AV et al. Association of Leisure Time Physical Activity Types and Risks of All-Cause, Cardiovascular, and Cancer Mortality Among Older Adults. JAMA Network Open. 2022. PMC9403775 ↗, Complementary study on 272,550 older adults.
Frequently asked questions
How many different physical activities should I practice to benefit from the "variety" effect?
The study doesn't establish a precise threshold, but participants in the most diversified quintile regularly engaged in several types of activities above a minimum engagement threshold. The Nurses' Health Study covered up to 11 activities, the Health Professionals Follow-Up Study up to 13. A pragmatic interpretation: combining three or four different categories (e.g., daily walking + strength training + a team or racket sport) already places you well above average.
Why does swimming show no clear benefit?
The study reports a hazard ratio of 1.01 (95% CI: 0.97–1.05) for swimming, meaning no statistically significant association with lower mortality. The authors don't settle the question definitively, but note that swimming was much less common in these cohorts than walking or running, reducing statistical power. Other studies have reported swimming benefits. It would be wrong to conclude that swimming is useless, the cautious interpretation is that this particular dataset doesn't capture a strong signal.
If I already walk a lot, is doing more really useful?
The study highlights a non-linear dose-response relationship: most of the benefit appears at the first engagement levels, then the additional effect declines. For many activities, doubling volume above the recommended thresholds (about 7.5–15 MET-hours per week) yields only marginal gains. Practical takeaway: if you're already at a comfortable volume, adding another activity type (strength training, cycling, racket sport) is likely more useful than further increasing walking time.
Does the study prove that variety causes a longer life?
No. This is a prospective observational study: it establishes a robust association between activity variety and lower mortality, after adjusting for total volume and many confounders (age, BMI, diet, smoking, etc.). But association is not causation. Variety could still be a marker of broader healthy behaviors (curiosity, resources, social support) rather than a direct causal factor. Demonstrating causation would require randomized trials over decades, practically unrealistic.
Was physical activity measured objectively (smartwatch, accelerometer)?
No. Physical activity was self-reported via questionnaire every two years. As Prof. Tom Yates (University of Leicester) emphasizes, self-reporting introduces biases that underestimate true associations. Studies using accelerometers (UK Biobank, NHANES) suggest all-cause mortality risk can be halved between very active and very inactive populations, far more than the figures reported here. The direction of effect is the same, the magnitude is likely underestimated.
Does this study replace official guidelines (WHO, CDC) on physical activity?
No, it complements them. Current WHO recommendations (150–300 min/week of moderate or 75–150 min/week of vigorous activity, plus two strength sessions) remain valid and well-founded. This study adds a dimension not made explicit in those guidelines: at constant total time, diversity of activities appears to bring an additional benefit. It's a refinement, not a replacement.
