VO2 max is fitness culture's favourite number. Four 2026 studies test what it establishes.

VO2 max has become the number consumer fitness culture organises itself around. The standing objection to the evidence behind it is confounding: fit people differ from unfit people in many ways a cohort study cannot fully adjust for.

Four studies published in 2026 attack that objection from different directions — a sibling design, a genetic design, a change-over-time design, and a molecular one. They largely agree, and where they disagree, they disagree in the same place.

The sibling test

The hardest confounder to remove is family. Shared genetics, childhood environment, parental behaviour and socioeconomic position all shape both adolescent fitness and adult disease.

A study in Circulation addressed that with full-sibling comparisons, using Swedish men who completed cardiorespiratory fitness testing at mandatory military conscription examinations between 1972 and 1995, followed in national registers until 31 December 2023 [s2]. The cohort comprised 1,124,049 men with a mean age of 18.3 years at testing [s2]. Over follow-up, 45,179 (4.0%) had a first atrial fibrillation event and 96,404 (8.6%) a first non-AF cardiovascular event, at median ages of 54.8 and 54.4 years [s2].

The design targets a real controversy: whether high fitness in youth raises the risk of atrial fibrillation later, a concern raised by observations in endurance athletes.

In the population-wide analysis controlling only for measured confounders, comparing the highest fitness decile with the lowest, there was a small excess of AF that exceeded the reduction in non-AF cardiovascular disease during early adulthood, with the reduction in non-AF disease becoming larger from age 45 onward [s2].

In the full-sibling comparisons, that age-dependent trade-off disappeared entirely, leaving no age window with a net cardiovascular disadvantage [s2]. From age 35, the reduction in non-AF cardiovascular disease was already larger (risk difference −0.11%, 95% CI −0.21% to −0.01%) than the excess in AF (RD 0.06%, 95% CI −0.01% to 0.12%) [s2]. By age 65 the gap had widened considerably: a reduction in non-AF cardiovascular disease of −3.91% (−5.40% to −2.42%) against an AF excess of 2.30% (1.15% to 3.45%) [s2].

Two things are true in that result at once. The AF excess is real and persists after sibling control. And it is outweighed, at every age examined, by the reduction in everything else.

The genetic test

A phenome-wide Mendelian randomisation study in Medicine & Science in Sports & Exercise used genetically predicted aerobic fitness as the exposure and screened 712 health-related phenotypes, using European-ancestry GWAS summary statistics, with an independent GWAS selected for validation of each association [s3].

Of 108 associations identified in discovery, 34 remained valid and statistically significant after validation [s3].

Higher genetically determined aerobic fitness was associated with lower lacunar stroke risk, lower arterial stiffness, higher heart rate variability, lower diastolic blood pressure, more favourable anthropometric measures, lower use of antidiabetic drugs, lower asthma risk, lower C-reactive protein, higher bone mineral density, favourable liver function and platelet-related traits, multiple haematological indices, and more years of schooling [s3].

Adverse associations were confined to atrial fibrillation, valvular heart disease and systolic blood pressure [s3].

That is the striking convergence. An analysis using genetic instruments, on entirely different data from the Swedish conscript registers, isolates the same exception — atrial fibrillation — from the same broad pattern of benefit.

The caution belongs on the record: Mendelian randomisation depends on assumptions about the genetic instruments that cannot be fully verified, and the analysis is restricted to European-ancestry data [s3]. The authors report directional concordance between discovery and validation, no evidence of horizontal pleiotropy across four methods, and a negative control analysis on hair colour [s3].

Change matters, not just level

Both of the above concern fitness as a fixed attribute. A study in the European Journal of Preventive Cardiology asked whether change in fitness predicts outcomes beyond baseline fitness [s1].

It analysed 11,530 UK Biobank participants aged 40 to 73 who completed two fitness assessments by submaximal cycle ergometry, calculating annualised change in estimated VO2 max [s1]. After excluding participants with prior cardiovascular disease or events within two years of follow-up, and over a median 14.0 years of follow-up (IQR 13.0 to 14.6), there were 921 incident cardiovascular events and 119 cardiovascular deaths [s1].

Models adjusted for baseline fitness, sociodemographic and lifestyle behaviours, genetic predisposition to cardiovascular disease, BMI change and physical activity [s1].

Each 1 mL·kg⁻¹·min⁻¹ per year improvement in VO2 max was associated with a hazard ratio of 0.85 (95% CI 0.75 to 0.96) for incident cardiovascular disease and 0.66 (0.51 to 0.87) for cardiovascular mortality [s1]. Declines were associated with increased risk, and the dose-response was approximately linear across the spectrum of change, without evidence of threshold effects [s1].

The authors' framing is that an annual improvement as little as 1 mL/kg/min may be enough to lower cardiovascular risk, independent of baseline fitness [s1].

That framing is the more actionable one, because it concerns a quantity a person can still move. The association held independent of baseline fitness [s1] — the starting point did not have to be high for the change to matter.

What fitness looks like in blood

A fourth study identified metabolomic and proteomic signatures of fitness in UK Biobank participants who completed a risk-stratified submaximal cycle ergometer test, then validated them in independent samples with metabolomics (n = 354,222) and proteomics (n = 29,961) [s4].

Higher fitness was characterised by downregulation of pathways related to inflammation, triglyceride metabolism, glycolysis and vascular dysfunction, and upregulation of pathways related to cholesterol transport, apolipoprotein particle size and cytoskeletal remodelling [s4]. The two signatures reflected fitness with R² between 0.50 and 0.60 [s4].

Over an average of 9 years of follow-up, with 27,659 all-cause deaths, the metabolomic signature was associated with 39% to 54% lower all-cause mortality risk across the discovery and validation cohorts, and markedly lower risk of type 2 diabetes (90% in both), cardiovascular disease (42% to 47%) and colorectal cancer (33% to 39%) [s4]. The proteomic signature was associated with 17% lower all-cause mortality and 22% to 39% lower risk of type 2 diabetes and cardiovascular disease [s4].

A caveat sits inside those figures. A blood signature trained to predict fitness will also capture inflammation, adiposity and metabolic health — conditions that both lower fitness and independently raise mortality. The signature's association with death is therefore stronger than fitness itself, not evidence that fitness is stronger than it appeared.

What all four support

Taken together: the association between cardiorespiratory fitness and cardiovascular outcomes survives sibling control, survives genetic instrumentation, and extends to change over time as well as level. The one consistent adverse signal, appearing in two independent designs, is atrial fibrillation — and in the sibling analysis it was outweighed at every age examined [s2].

None of this makes a VO2 max estimate from a wrist device a clinical measurement. All four studies relied on supervised fitness testing [s1][s2][s4] or on genetic proxies [s3]. What the number means depends entirely on how it was obtained.

This article is informational and is not medical advice.