Lifting weights tracked with lower mortality, and the benefit stopped at two hours a week

An analysis published in the British Journal of Sports Medicine measured how much people lifted not once but repeatedly — by questionnaire at baseline and every two years afterwards, for as long as three decades [s1].

The cohorts

The authors pooled participants from three long-running prospective cohorts: the Health Professionals Follow-up Study (1992–2022), the Nurses' Health Study (2002–2021) and the Nurses' Health Study II (2003–2021) [s1]. Weekly resistance training duration and aerobic exercise duration were assessed by validated questionnaire at baseline and biennially afterwards [s1].

That amounted to 147,374 participants — 31,540 men and 115,834 women — followed for up to 30 years, during which 35,798 deaths were documented [s1].

Repeated measurement matters more here than the sample size. A single baseline questionnaire cannot distinguish someone who lifted for one year from someone who lifted for twenty. Biennial updating tracks the exposure as it changes.

The dose-response curve

Compared with no resistance training, 90 to 119 minutes per week was associated with [s1]:

• 13% lower all-cause mortality (HR 0.87, 95% CI 0.81 to 0.95)
• 19% lower cardiovascular mortality (HR 0.81, 0.67 to 0.97)
• 27% lower neurological disease mortality (HR 0.73, 0.58 to 0.92)

All adjusting for aerobic activity [s1].

Above 120 minutes per week, no additional benefit was observed [s1]. The curve flattens. Whatever the mechanism, it appears to saturate at roughly two hours of resistance work weekly — well below what most strength programmes prescribe, and well below what many people who lift actually do.

The cancer result runs the other way

The cancer mortality finding is the one that resists a simple reading. Reduced cancer mortality risk was seen only at lower levels of resistance training: HR 0.91 (0.86 to 0.97) for 1–29 minutes per week, and HR 0.88 (0.81 to 0.97) for 30–59 minutes per week [s1].

At higher volumes, the association was not observed. That pattern is difficult to explain mechanistically and easy to explain statistically — small numbers of high-volume lifters, imprecise self-report at the top of the range, and the ordinary noise of subgroup estimates in an observational cohort. It should be read as an inconsistency in the data, not a warning about lifting more.

The joint analysis with aerobic activity

The more practically interesting result is what happens when resistance and aerobic training are considered together.

Using as reference those with inadequate aerobic activity (under 7.5 MET-hours per week) and no resistance training, mortality risk was lowest in two groups [s1]. The first combined high aerobic and resistance training — for example, HR 0.55 (0.50 to 0.60) for 30 to under 45 MET-hours per week of aerobic activity together with 60–119 minutes per week of resistance training [s1]. The second was anyone reaching 45 MET-hours per week or more of aerobic activity, regardless of resistance training level, with hazard ratios from 0.53 to 0.58 [s1].

Read carefully, that is two distinct findings. Resistance training was associated with further reduced mortality risk at all levels of aerobic activity up to around 45 MET-hours per week [s1] — so for most people, adding lifting adds something. But at the very top of the aerobic range, the resistance-training gradient no longer separated the groups.

What an observational cohort cannot do

Nobody was randomised. People who lift weights consistently for thirty years differ from people who do not in ways no questionnaire captures — baseline health, injury history, income, the absence of conditions that make lifting impossible. Reverse causation runs in the obvious direction: illness stops people training long before it kills them.

The design mitigates some of this. Repeated exposure measurement over 30 years is a stronger instrument than a baseline snapshot, and adjustment for aerobic activity addresses the most obvious confounder — that lifters also run.

The populations are also specific. These are US health professionals and nurses, predominantly women in two of the three cohorts, with above-average health literacy and healthcare access. The hazard ratios describe them.

What it changes

The plateau is the finding with practical weight. The evidence assembled here associates the bulk of the mortality signal with a volume most people could reach in two or three sessions a week, and does not associate additional benefit with more.

That is a claim about mortality, and only mortality. Strength, muscle mass, bone density and performance have their own dose-response curves, and this study did not measure any of them.

This article is informational and is not medical advice.