Cardiorespiratory Fitness and Mortality: The Dose-Response

Cardiorespiratory fitness predicts survival more consistently than most people's preventive-care checklist suggests. In one clinically referred cohort, low fitness accounted for an estimated 12.9% of all-cause mortality at the population level, compared with 8.8% for self-reported physical inactivity. The result does not mean fitness causes every one of those deaths. It does mean a VO2max or exercise-capacity estimate can contain information that blood pressure, lipids, glucose, body weight, and an activity questionnaire miss.

The evidence base is unusually large for an exercise-related risk marker. A 2024 overview of 26 meta-analyses, covering 199 unique cohort studies and more than 20.9 million observations, found that cardiorespiratory fitness (CRF) consistently predicted lower all-cause, cardiovascular, and cancer mortality. A separate 2025 meta-analysis of 42 studies, 35 cohorts, and 3.8 million observations reached the same broad conclusion. The important caveat comes first: these are mainly observational data. Fitness may partly capture lifelong health, socioeconomic conditions, underlying disease, access to care, and genetics. It is a powerful prognostic signal; it is not a randomized proof that raising VO2max by a given amount prevents a specified number of deaths.

The mortality gradient starts steep

CRF describes the body's capacity to deliver and use oxygen during sustained exercise. It is usually expressed as VO2max or VO2peak, often converted into metabolic equivalents (METs). One MET is roughly 3.5 mL of oxygen per kilogram of body weight per minute.

The association is graded rather than binary. In a 2021 prospective cohort study by Myers et al., 5,890 male veterans referred for clinical exercise testing were followed for an average of 9.9 years; 2,728 died. After adjustment for clinical and demographic factors, every additional MET of exercise capacity was associated with a hazard ratio for all-cause mortality of 0.85 (95% CI, 0.83–0.88; P<.001). That corresponds to about 15% lower relative mortality risk per MET in that model.

The gap between the cohort's extremes was larger still. Participants in the least-fit quartile had an all-cause mortality hazard ratio of 5.99 (95% CI, 4.9–7.3) compared with the most-fit group. That is an association within a selected clinical population, not a promise that training will turn an individual's risk into one-sixth of its former value. Still, it gives the size of the observed gradient some teeth.

Across meta-analyses, the recurring estimate is about 13% lower all-cause mortality and 15% lower cardiovascular-event risk per 1-MET increment. Han et al.'s 2022 dose-response meta-analysis found a nonlinear relationship for all-cause, cardiovascular, and cancer mortality: the curve drops most sharply when people move from very low fitness into a moderate range. Elite fitness still associates with lower risk, but the first gains out of the basement appear to deliver the largest return.

The practical target is not an endurance-sport identity. It is avoiding the very-low-fitness range, where the observed mortality curve is steepest.

Why fitness can outperform an activity questionnaire

Physical activity and fitness overlap, but they are not interchangeable. An activity questionnaire asks what a person reports doing. CRF records the integrated result of activity, cardiovascular function, lung function, body composition, illness, medication effects, and inherited physiology. That may explain why low CRF had a higher population-attributable risk than self-reported inactivity in the Myers cohort.

That same study estimated an exposure impact number (EIN) of 3.8 for low CRF below 5.0 METs (95% CI, 3.4–4.3; P<.001), lower than the EINs reported there for diabetes, smoking, hypertension, and physical inactivity. EIN is a clinical-efficiency estimate: the number of people who would need to be addressed through a risk factor to avert one death under the study's assumptions. It is a striking figure, but it deserves a warning label. The comparison comes from one cohort of male veterans referred for exercise testing, not a representative sample of women or healthy adults. It also does not establish that fitness improvement alone would produce that mortality benefit.

Lang et al.'s 2024 overview reports that adding CRF can improve mortality risk reclassification beyond conventional risk factors. The American Heart Association has recommended estimating CRF at least annually during routine healthcare examinations, yet it is still rarely part of ordinary visits. A formal treadmill test for everyone would be impractical. An estimate based on validated submaximal testing may still be useful for tracking, while direct cardiopulmonary exercise testing remains more precise.

The signal survives BMI and established disease

The fitness association does not disappear when body size enters the model. A 2025 systematic review and meta-analysis by Weeldreyer et al. found that CRF independently predicted all-cause and cardiovascular mortality across BMI categories. This does not make excess adiposity harmless, nor does it settle every causal pathway. It means BMI alone is a poor substitute for measuring aerobic capacity; a person can have a higher BMI and higher fitness, or a normal BMI and very low fitness.

The relationship also appears after cardiovascular disease has already been diagnosed. In a meta-analysis by Ezzatvar et al., 21 prospective studies included 159,352 patients with established cardiovascular disease, 38.1% of them women. Participants in the highest CRF category had lower all-cause mortality than those in the lowest category (pooled HR, 0.42; 95% CI, 0.28–0.61) and lower cardiovascular mortality (HR, 0.27; 95% CI, 0.16–0.48). Because these are observational comparisons among patients, they support risk stratification rather than a claim that fitness testing or exercise alone caused those differences.

Age does not erase the signal. In a Norwegian cohort of 788 women and 777 men aged 70 to 77 years, Tari et al. linked higher VO2peak with lower five-year all-cause mortality. Compared with unfit participants, the reported hazard ratios for moderate-to-higher fitness were 0.34 (95% CI, 0.15–0.78) in men and 0.41 (95% CI, 0.17–0.98) in women. Both confidence intervals exclude 1.0, although the female estimate is imprecise because few deaths occurred.

Biology supports the pattern, but does not prove the mechanism

Fitness has plausible pathways to cardiovascular survival. In a cohort of 30,077 adults, 31.3% women, followed from 2000 through 2016, Farrell et al. found the best survival pattern among participants with high CRF and low high-sensitivity C-reactive protein; low CRF plus elevated CRP had the worst pattern. The study adjusted for several standard risk factors, including BMI, blood pressure, lipids, glucose, smoking, age, and sex. CRP is an inflammation marker, so the finding is compatible with shared inflammatory pathways. It cannot show that reduced inflammation caused the fitness-mortality association.

Vascular findings point in the same direction without settling causality. Fleenor et al. found higher CRF associated with markers of healthy vascular aging in 101 apparently healthy adults aged 50 and older. Better endothelial function and less arterial stiffness could help explain lower cardiovascular risk, but this small observational study measured markers, not deaths.

What the headline leaves out

The most dramatic population-attributable and EIN estimates have important overlap problems. Diabetes can impair CRF, as a 2024 systematic review by Alvares et al. found for both type 1 and type 2 diabetes. Some mortality attributed to low fitness in a population-attributable analysis may therefore be shared with diabetes-related risk. A Finnish prospective cohort found VO2max predicted cardiovascular mortality independently of glucose status, which supports separate predictive value, but it does not quantify how much of the 12.9% population-attributable estimate is independent of diabetes.

Genes also influence CRF. Klevjer et al.'s genome-wide association study identified genetic determinants of fitness, so the population fraction associated with low CRF is not automatically the fraction that training can change. Measurement varies across studies as well: treadmill tests, cycle tests, and estimated VO2max do not carry identical predictive accuracy. The 2025 Singh meta-analysis found directly measured CRF predicted mortality better than estimated CRF.

Finally, movement at work is not a reliable stand-in for aerobic conditioning. Ketels et al. reported that occupational physical activity did not provide the same CRF and mortality pattern as leisure-time aerobic training. Heavy work may impose repeated physical strain without the recovery and progression that build aerobic capacity.

The useful conclusion

CRF belongs in a serious risk assessment. The observed mortality gradient is large, repeats across many cohorts, persists across BMI groups and clinical settings, and appears in older adults. The strongest evidence supports measurement and risk stratification, especially for people at the low end of fitness. It does not justify treating a wearable's VO2max estimate as a death forecast or claiming that one training plan cancels smoking, diabetes, hypertension, or excess adiposity.

If your fitness is unknown, estimate it with a validated submaximal protocol or discuss formal testing with a clinician when appropriate. Then use the number as a baseline for progressive aerobic training and follow-up, not as a personality test. The most defensible goal is simple: move out of the low-fitness range and keep the trend moving upward.