Check · Fitness · VO₂max In review

Is improving your VO₂max one of the most powerful things you can do to live longer?

Claim attributed to Peter Attia , Physician and author of the #1 bestseller Outlive, in which VO₂max and zone-2/zone-5 training are central; runs a paid membership where the detailed VO₂max material is paywalled. In his own writing he is careful, he frames the data as an observational "association" and a "monotonic relationship between fitness and mortality." We grade the popularized, strongly causal version of the claim ("one of the most powerful things you can do to extend your lifespan"), not his hedged wording.

Verdict Mixed

Evidence grade B Moderate certainty

As a mortality marker, fitness is one of the strongest we have, that part is real. But the leap from "fit people die less" to "raising your VO₂max number is one of the most powerful ways to live longer" outruns the evidence: the one direct causal test was null for longevity, and randomized exercise has never reliably cut all-cause death.

Fitness is one of the strongest things you can measure to predict death, but the one causal test found raising the number doesn't extend life, and randomized exercise has never reliably cut all-cause mortality.

The theory

What it’s supposed to target

Cardiorespiratory fitness
Mitochondrial density / function
Cardiac output
Oxygen delivery and use

VO2max is the most oxygen your body can take in and use, set by how well the heart pumps, blood carries oxygen, and mitochondria in muscle burn it. The longevity logic is direct: a higher ceiling reflects a stronger cardiovascular system and denser, healthier mitochondria, and it buys a larger reserve above the threshold of frailty as you age.

This is the unusual entry where mechanism and evidence line up: large cohort studies consistently show higher cardiorespiratory fitness tracks with much lower mortality, with little sign of a downside. The fair caveats are that most data are observational (fit people differ in many ways) and that VO2max is partly genetic, so it is not fully under your control. A sound mechanism with unusually strong, though still associational, human data.

Mechanism is theory, not proof. A plausible pathway explains why something might work, not whether it does. The verdict rests on the evidence below, not the elegance of the theory.

The claim

What would have to be true

The claim rests on a chain of four links, and they do not all hold:

1. VO₂max strongly predicts who lives and who dies. Holds, robustly. This is among the best-replicated findings in the field.

2. Raising your own VO₂max therefore lowers your own risk by something like the observed gap. Does not follow. The steep observational gradient is inflated by reverse causation (sick people have low fitness) and confounding, and the one causal test found no effect on longevity.

3. A randomized intervention that raises fitness produces longer life. Not shown. The closest evidence, exercise-based cardiac rehab, does not significantly reduce all-cause mortality, though it does cut cardiovascular death long-term.

4. It is one of the MOST powerful levers available. Untested as a comparison. No study ranks modifiable interventions head-to-head on lifespan.

The evidence

What the evidence actually shows

As a marker, fitness is about as strong as it gets

This is the half of the claim that holds up, and it holds up well. In the largest single study, 122,007 adults followed a median of 8.4 years at the Cleveland Clinic, all-cause mortality fell monotonically with cardiorespiratory fitness, with the authors noting no plateau and no U-shape: the fittest kept doing better, all the way up. The elite-versus-low-fitness adjusted hazard ratio was 0.20 (95% CI 0.16–0.24), an 80% lower risk [1].

The magnitude genuinely rivals classic risk factors. In that same cohort, being in the lowest-fitness group carried a mortality risk comparable to or greater than coronary artery disease (HR 1.29), smoking (HR 1.41) and diabetes (HR 1.40) [1]. A JAMA meta-analysis of 33 cohorts (~103,000 people) put a number on the dose-response: each 1-MET higher fitness carried an all-cause mortality relative risk of 0.87 (0.84–0.90) [2]. The American Heart Association reviewed this body of work and concluded low fitness is "a potentially stronger predictor of mortality than established risk factors such as smoking, hypertension, high cholesterol, and type 2 diabetes," recommending it be treated as a clinical vital sign [3].

If the claim were only "VO₂max is one of the most powerful things you can *measure* to gauge mortality risk," it would earn a clean Supported. The trouble is the word the popular version smuggles in: *do*.

The one direct causal test was null for longevity

A predictor is not a lever. The cleanest available test of whether the VO₂max number itself causes longer life is Mendelian randomization, using genetic variants as nature's randomized trial, which sidesteps the reverse-causation and lifestyle confounding that plague cohorts. A 2-sample study used genetic instruments for VO₂max (UK Biobank, ~70,000) against longevity data on ~1,012,240 parents and concluded, verbatim: "Despite being a strong predictor of mortality, VO₂max is not causally associated with T2D or longevity" [6].

What makes this more than an underpowered miss is what the same instruments *did* detect. They found body-fat percentage, lean mass, physical activity and hematocrit causally *raise* VO₂max, and that body-fat percentage and type 2 diabetes causally affect longevity, while the VO₂max-to-longevity path was null. That is an internally coherent picture in which VO₂max is a downstream readout of underlying health, not the upstream cause of longer life [6]. The genetic tools worked; they simply found the lever elsewhere.

One honest caveat on the caveat: the VO₂max instrument is built from *estimated* fitness via a submaximal cycle test, there is some sample overlap among the genetic datasets, and parental lifespan is a noisy proxy, so the null is not bulletproof. It does not *prove* VO₂max is causally inert. But because the same instruments detected the expected upstream effects, the result cuts against, rather than merely fails to support, the simple causal story.

Randomized exercise has not reliably cut all-cause death

The other way to test causation is to actually intervene. No trial has ever raised healthy people's VO₂max and then counted deaths over decades, that study is essentially unrunnable (more on why below). The closest randomized evidence is exercise-based cardiac rehabilitation, which reliably raises VO₂max. A Cochrane review of 85 trials and 23,430 coronary-disease patients found all-cause mortality RR 0.87 (0.73–1.04) short-term and 0.91 (0.75–1.10) long-term, both crossing 1.0, neither statistically significant [7]. Even cardiovascular mortality up to 12 months was non-significant; the substantial cardiovascular-mortality benefit (RR 0.58, 0.43–0.78) and the drop in hospitalizations emerge only over the longer term [7].

So randomized exercise that demonstrably raises fitness delivers, at best, a modest and frequently non-significant all-cause mortality effect, nowhere near the 80% observational gap. A separate critical review of the RCT evidence went further, concluding flatly that "the current evidence shows that exercise does not prevent premature mortality or CVD" (it prevents type 2 diabetes only when combined with dietary change) [8]. That is a real, published, disconfirming finding on the lifespan question, not merely a note that good trials are missing.

Two clarifications keep this fair. First, cardiac-rehab patients are not healthy adults, so this transfers imperfectly. Second, a non-significant result is not proof of no effect. But it is the best randomized signal we have, and it does not support the strong causal magnitude the claim implies.

Why the observational gap overstates the achievable benefit

The 80% figure is real but treacherous. Sick, frail, sedentary and pre-diseased people have low VO₂max for reasons, subclinical heart disease, undiagnosed cancer, frailty, that independently raise their mortality. This is reverse causation, and it inflates the apparent benefit of "being fit" because low fitness is partly a *symptom* of illness, not only a cause of it.

The authors of the anchor study say so themselves. Mandsager and colleagues write that "the association between CRF and mortality does not prove causation" and that they cannot discern whether high fitness causes better outcomes or merely preselects already-healthier people [1]. Cohorts try to mitigate this, excluding early deaths, adjusting for known confounders, but cannot eliminate it. The honest reading is that intervening on your own VO₂max almost certainly buys *less* than the elite-versus-low contrast advertises.

Finally, the superlative. "One of the *most* powerful things you can do" is a ranking claim, and no study has ever pitted raising VO₂max head-to-head against not smoking or controlling blood pressure on a lifespan endpoint. Observational predictor rankings are themselves confounded and do not establish that *acting* on fitness beats acting on those other levers [7,8].

Evidence quality

Studies, graded, and who paid

VO₂max is a powerful predictor of all-cause mortality A High certainty

Large, consistent, dose-dependent across cohorts of ~100,000+; endorsed by the American Heart Association. This half is not in dispute.

Exercise that raises VO₂max is good for cardiovascular health B Moderate certainty

Randomized cardiac-rehab trials cut cardiovascular mortality and hospitalizations long-term, a real, if disease-specific, benefit.

Raising your own VO₂max number causally extends lifespan C Low certainty

The one genetic causal test was null for longevity; the best exercise RCTs show no significant all-cause mortality reduction. Tested, and weaker than the marker implies.

It is one of the MOST powerful things you can do, a comparative superlative D Very low certainty

Never head-to-head tested against smoking cessation or blood-pressure control on a lifespan endpoint. Asserted, not demonstrated.

Cited studies with type, size, funding/conflicts, and limitations.
#	Study	Type	Size	Funding / COI	Key limitations
1	Mandsager et al., JAMA Network Open, 2018	Retrospective observational cohort	n=122,007; median 8.4-yr follow-up; 13,637 deaths	Independent Conflicts of interest: "None reported." Single academic center (Cleveland Clinic); no industry sponsor. Authors explicitly state the association "does not prove causation."	Observational; cannot rule out reverse causation/preselection (authors say so). Population referred for treadmill testing, not a general-population sample.
2	Kodama et al., JAMA, 2009 (meta-analysis)	Meta-analysis of 33 observational cohorts	~102,980 participants; 6,910 deaths (all-cause analysis)	Independent Listed as "Research Support, Non-U.S. Gov't." No commercial product sponsor.	Pools observational cohorts only, establishes a consistent dose-response association, not causation.
3	Ross et al., AHA Scientific Statement, Circulation, 2016	Scientific statement / expert consensus	Synthesis of observational evidence (no single sample)	Independent American Heart Association statement; indexed record lists two VA grants (I01 RX000344, IK6 RX002477). No commercial product sponsor.	Synthesizes observational data; explicitly frames fitness as a predictor/vital sign, not a proven causal intervention.
4	Lee et al., Aerobics Center Longitudinal Study, Circulation, 2011	Prospective observational cohort	n=14,345 men; ~11.4-yr follow-up; 914 all-cause / 300 CVD deaths	Mixed NIH grants (AG06945, HL62508, DK088195) PLUS "an unrestricted research grant from The Coca-Cola Company." Authors declared no conflicts. The Coca-Cola tie is flagged given the company's documented interest in framing inactivity, rather than diet, as the driver of poor health, an interest this study's pro-fitness conclusion happens to serve.	Observational; men only; referral/treadmill-tested population; the change-in-fitness inference rests on this single-sex cohort. Weakest of the anchor studies, should not carry the "improving fitness" claim alone.
5	Kokkinos et al., JACC, 2023 (VA Exercise Testing cohort)	Retrospective observational cohort	n=93,060 veterans; two treadmill tests ~5.8 yr apart; median 6.3-yr follow-up	Independent Veterans Affairs / academic. Authors reported "no relationships relevant to the contents of this paper to disclose." No commercial sponsor identified.	Observational; veteran population skews male/middle-aged; a ≥2-MET fitness decline tracked with 69–74% higher mortality, but change-in-fitness associations remain susceptible to reverse causation.
6	Kjaergaard et al., J Clin Endocrinol Metab, 2025 (Mendelian randomization)	2-sample Mendelian randomization (genetic causal inference)	VO₂max instrument ~70,000 (UK Biobank); longevity ~1,012,240 parents	Independent Novo Nordisk Foundation, NIH NIDDK (R01 DK124258, R01 DK129850), Boston Children's Hospital endowment. "The authors have nothing to disclose." Transparency note: the Novo Nordisk Foundation is a pharma-linked philanthropy; the null finding here does not favor any product, so the conflict risk is low.	VO₂max instrument uses estimated fitness (submaximal test); some GWAS sample overlap; parental lifespan is a noisy longevity proxy. The null is not definitive, but the same instruments detected expected upstream effects, so it is a genuine signal, not a dead instrument.
7	Dibben et al., Cochrane Review, 2021 (exercise-based cardiac rehab)	Cochrane systematic review / meta-analysis of RCTs	85 randomized trials; 23,430 coronary heart disease patients	Independent Cochrane Collaboration review; standard Cochrane independence/COI policy.	All-cause mortality non-significant (RR 0.87 short-term, 0.91 long-term; CIs cross 1.0; moderate certainty, downgraded for imprecision). Population is coronary-disease patients, not healthy adults, transfers imperfectly to the general claim.
8	Ballin & Nordström, J Intern Med, 2021 (critical review of RCTs)	Critical review of randomized-trial evidence	Narrative synthesis of long-term exercise RCTs	Independent Academic (Umeå University); no commercial sponsor identified.	Narrative, not a pooled meta-analysis. Concludes RCT evidence "shows that exercise does not prevent premature mortality or CVD", an active negative finding, weighed alongside the structural reasons large mortality-powered exercise trials are rarely run.

The split is between marker and lever, and the studies fall cleanly on either side. Everything that measures *association*, Mandsager, Kodama, the AHA statement, the VA cohort, is large, consistent and points the same way: fit people die less, in a tidy dose-response. Everything that tests *causation*, the Mendelian-randomization study and the Cochrane RCT synthesis, comes back null or non-significant for the outcome that matters, longevity and all-cause death. The strength of the claim's marker half is precisely what makes the weakness of its causal half easy to miss.

The science is largely independent of the claimant, which is a point in the claim's favor. Unlike a typical supplement check, the anchor mortality evidence here is not sponsor-funded and predates the person popularizing it. The one funding flag worth naming is the Coca-Cola unrestricted grant behind the weakest anchor (Lee 2011), a funder with a documented interest in blaming inactivity over diet, and the study most often leaned on for the "get fitter" inference [4]. The decisive null result (the MR study) is funded by academic and NIH grants plus a pharma-linked foundation that has no product riding on the answer [6].

See who funded the studies behind every Caveat check

Stay neutral

Unproven ≠ disproven

Unproven is not disproven, but here, part of the claim has actually been tested. The narrow, untested question is the ideal one: randomize healthy adults to raise VO₂max, then follow them for decades and count deaths. That trial does not exist, so the lifespan endpoint of a VO₂max-raising intervention is genuinely *unproven*. But the broader causal magnitude is not merely untested: a direct genetic causal test was null for longevity [6], and the best randomized exercise synthesis was non-significant for all-cause death [7]. So this is not a case of pure absence of evidence, there is disconfirming evidence on the causal *size*, which is why the verdict is Mixed rather than Unproven.

The decisive trial is also structurally near-impossible, which is worth stating fairly. Randomizing thousands of people to either chase fitness or deliberately stay unfit, then following them 15–30 years, runs into an ethically dubious control arm, near-certain contamination (the unfit controls start exercising), dropout, and a cost few funders will bear. So we will likely never get clean causal proof either way. That argues for humility, not cynicism: the biology is plausible, the marker is powerful, and exercise has well-established benefits short of the lifespan endpoint.

The gap

Where claim and evidence diverge

The gap is the swap of a predictor for a lever, and of a measured fact for a comparative superlative. "VO₂max strongly predicts mortality" (true, grade A) is quietly upgraded to "raising your VO₂max is one of the *most powerful things you can do* to live longer" (a causal ranking claim that the null genetic test, the non-significant RCTs, and the complete absence of any head-to-head comparison do not support).

The honest reframing: improving your fitness is well-evidenced as good for your cardiovascular health and almost certainly worthwhile, the marker is real, the mechanism is plausible, exercise cuts cardiovascular events. What is *not* established is that the VO₂max number itself is a uniquely powerful lifespan dial, or that pushing it up delivers anything close to the 80% mortality gap the cohorts display.

Follow the funding

The money trail

Peter Attia profits directly from Outlive, a #1 New York Times bestseller in which VO₂max and zone-2/zone-5 training are central, and from a paid subscription membership where the detailed VO₂max content sits behind a paywall.

Per his own disclosures, Attia holds equity in numerous health and wellness companies, including Oura Health (the Oura ring, which markets fitness and cardiovascular tracking), AG1 (where he is also a scientific advisor), LMNT, David Protein, and others, a financial interest in the broader "measure and optimize your fitness" ecosystem.

To his credit, and verifiably on the same disclosures page, Attia states he does not do affiliate or kickback marketing and has never taken an affiliate payment for products he discusses, which distinguishes him from typical supplement marketers.

The point that matters most for this check: none of the anchor VO₂max-mortality studies are funded by Attia or by anything he sells. The fitness-mortality literature predates him and is independent of him. His incentive is to popularize the strong, motivating version of the claim, but the evidence the claim rests on is not his. We state these as facts for the reader to weigh, not as an accusation.

Bottom line

The honest read

This is one of the better cases in longevity, which is exactly why the distinction matters. Cardiorespiratory fitness is among the most powerful mortality *predictors* ever measured, the cohorts are large, consistent, dose-dependent, and the American Heart Association treats fitness as a clinical vital sign. If the claim stopped at "your VO₂max tells you a great deal about your risk," it would be Supported without hesitation.

But the popular version says something stronger: that pushing the number up is one of the most powerful things you can *do* to extend your life. There the evidence thins. The one direct causal test was null for longevity and suggests fitness is a downstream readout of health rather than its cause; the best randomized exercise trials do not significantly reduce all-cause death; and the "most powerful" superlative has never been tested against smoking cessation or blood-pressure control. So: get fit, the marker is real, the mechanism is sound, and exercise has solid cardiovascular benefits. Just know that the 80% mortality gap is a property of fit *people*, not a promise of what raising your own number will buy.

Falsifiable

What would change this verdict

A long-term randomized trial that raises VO₂max in generally healthy adults and shows a significant reduction in all-cause mortality or hard clinical outcomes → moves toward Supported.

A Mendelian-randomization study using a higher-quality, directly-measured VO₂max instrument (and non-overlapping samples) that finds a causal effect on longevity → moves toward Supported and would overturn the current null.

Evidence that, head-to-head, intervening on fitness extends life by less than intervening on smoking or blood pressure → cements the superlative as Unsupported even if the directional benefit is real.

Receipts

Sources

Common questions