The Two-Hour Ceiling: What 30 Years of Data Says About How Much Strength Training Is Enough

I have lifted three to five days a week for years, and until a few weeks ago I could not have told you how many minutes that was.

Not roughly. Not at all. I could tell you my working weight on a trap-bar deadlift, my rep scheme, which days I press and which days I pull. I track my food to the gram. I know my fasting window to the quarter hour. But the total weekly duration of the thing I consider the foundation of the whole practice? Never measured it. It had never occurred to me that it was a number.

Then a study landed that made it one.

In June, a team at the Harvard T.H. Chan School of Public Health published an analysis in the British Journal of Sports Medicine that followed 147,374 adults for up to thirty years and asked a question the official guidelines have always ducked: not whether you should lift, but how much. The answer they landed on — roughly 90 to 120 minutes a week — went around the internet in about a day, usually with the word "optimal" attached to it.

So I got a notebook and reconstructed a normal week. Warm-up sets counted. Rest between sets counted, because that is how the questionnaires ask it. Fifty minutes a session. Three sessions in a lean week, five in a good one. Which puts me somewhere between 150 and 250 minutes — call it two and a half hours on my quietest week and just over four on my best.

So my floor is above the ceiling. Even the week where I feel like I am barely holding the practice together sits past the point where the study says the line goes flat, and my good weeks sit at roughly double it. I had assumed I would find myself somewhere near the sweet spot, quietly validated. Instead I found I had been paying into something for years at a rate the evidence does not support.

What follows is what the study actually found, what I changed because of it, and — the part almost nobody covering this got to — the several large things it cannot tell you.

What the Study Actually Found

Start with the design, because the design is most of the story.

This was an observational cohort analysis, not a trial. Nobody was assigned to lift. The researchers pooled three of the longest-running prospective cohorts in American epidemiology — the Health Professionals Follow-Up Study (31,540 men, 1992 to 2022) and the Nurses' Health Study and Nurses' Health Study II (115,834 women, running through 2021). Participants reported their weekly resistance training and aerobic minutes on validated questionnaires at baseline and then every two years for decades. Over that stretch, 35,798 of them died.

Comparing people by how much they lifted, and adjusting for how much aerobic activity they did, the researchers found this: at 90 to 119 minutes of resistance training a week, all-cause mortality was 13% lower than in people who did none (hazard ratio 0.87, 95% confidence interval 0.81 to 0.95). Cardiovascular mortality was 19% lower. Death from neurological disease was 27% lower.

And then the curve went flat. Above about 120 minutes a week, there was no further reduction. More lifting, same association.

That flattening is the whole reason this study travelled. It is a genuinely useful shape, because it argues against the thing most committed people quietly believe — that the relationship between effort and outcome keeps climbing as long as you keep climbing.

The curve rises, and then it stops rising. That shape is the finding. The number attached to it is the least reliable part.

The Number Is Not a Number

Here is what did not make the headlines.

In 2022, the same journal published a systematic review and meta-analysis by Momma and colleagues, pooling sixteen prospective cohort studies. Same question. It found muscle-strengthening activity associated with a 10 to 17% lower risk of all-cause mortality, cardiovascular disease, total cancer, diabetes and lung cancer — and it found a roughly J-shaped dose-response, with the strongest association at about 30 to 60 minutes a week, and no conclusive evidence of additional benefit beyond about an hour.

Thirty to sixty. Ninety to one-twenty. Two large, careful, peer-reviewed analyses in the same journal, disagreeing by a factor of two to three about where the sweet spot sits.

They are not really contradicting each other. They measured different populations with different instruments over different windows, and self-reported exercise duration is a soft measurement in every one of them. What replicates across both is the shape: a meaningful benefit at a modest dose, and then a flattening. What does not replicate is the number.

The lead author of the new paper said as much himself. Yiwen Zhang, the postdoctoral fellow who led the analysis, told reporters plainly: "Importantly, 90 to 120 minutes per week should not be interpreted as a strict target or threshold."

I want to sit on that for a second, because it is the single most ignored sentence in the entire news cycle. The person closest to the data went out of his way to say the number is not a target — and the coverage put the number in the headline anyway, usually with "optimal" in front of it.

What This Evidence Cannot Tell You

This is the section I would want to read first, so I am putting it before the practical part rather than burying it at the end.

It cannot tell you that lifting caused anyone to live longer. This is an association drawn from observation. Nobody randomised 147,000 people into a lifting arm and a no-lifting arm and waited thirty years. A 2022 systematic review in the American Journal of Preventive Medicine looking specifically at resistance training and mortality found eleven qualifying studies — one randomised trial and ten cohort studies. There is no adequately powered randomised trial with death as its primary endpoint, and there almost certainly never will be, because running one would cost a fortune and take a career.

So the confounding runs exactly where you would expect. People who lift twice a week also tend to sleep more, smoke less, drink less, sit less, and — crucially — be healthy enough to lift in the first place. Some of that 13% is the lifting. Some of it is everything that travels with a person who lifts. The researchers adjusted for a great deal, but adjustment is not randomisation, and residual confounding is the permanent tax on this kind of work.

It cannot tell you the shape of the training. The cohorts captured minutes. They did not capture load, intensity, proximity to failure, exercise selection, or how long anyone rested between sets. Ninety minutes of unhurried machine circuits and ninety minutes of heavy compound work with real rest intervals both landed in the same bucket. Anyone reading a prescription for how to train out of this paper is reading something that is not there.

It cannot tell you that more than two hours is wasteful, and it certainly cannot tell you it is harmful. A plateau is not a downturn. The line stopped going down; it did not turn back up. Those are different findings, and the difference matters if you happen to love training.

It cannot tell you this generalises cleanly to you. These were nurses and health professionals — better educated, more health-literate, and more consistent than the general population by a wide margin. That is what makes their thirty-year follow-up so good, and it is also what limits it.

And it cannot tell you why. There is a popular explanation making the rounds right now — muscle as an endocrine organ, releasing myokines that talk to the liver, the brain, the vasculature. The biology is real. But the chain from a myokine released on a Tuesday in 1998 to a death certificate in 2022 is not established, and this study did not test mechanism at all. It is a hypothesis wearing the clothes of an explanation.

Adjustment is not randomisation. Some of that 13% is the lifting, and some of it is everything that travels with a person who lifts.

The Ceiling You're Picturing Is a Different Ceiling

This is where I nearly made a real mistake, and where I suspect a lot of people will.

When I read "no additional benefit above 120 minutes," some part of my brain heard "the fourth session is pointless." It is not, and conflating those two claims is a category error worth naming clearly.

The mortality curve and the adaptation curve are two different curves answering two different questions.

A 2025 Bayesian meta-regression in Sports Medicine pooled the resistance training literature to ask how weekly volume relates to muscle and strength gains. The finding was unambiguous in direction: the probability that more weekly sets produce more hypertrophy and more strength was 100%. The slope never turned negative. There were clear diminishing returns — sharply so for strength — but the estimate at an average weekly volume of about twelve sets was still roughly a 0.24% increase in hypertrophy for each additional set.

So: more volume keeps building more muscle, with each set buying less than the one before it. Meanwhile, the association with dying later stops improving somewhere around two hours a week.

Both are true at once, because they are measuring different outcomes. If your goal is a bigger squat or more muscle on your frame at 54, volume above two hours still pays — just less per hour than it used to. If your goal is the mortality curve specifically, the evidence says you reached the interesting part of it a while ago.

Knowing which goal you are actually training for is the whole game. I have watched myself add a fourth session and quietly tell myself it was for my health, when it was really for my ego, or for the way a hard session settles my head on a bad week. Those are legitimate reasons. They are just not the reason I was claiming.

More volume still builds more muscle. It just stops buying more years. Know which one you're training for.

Strength Alone Wasn't the Winner

The finding I care about most got the least coverage, because it is not a headline number.

The lowest mortality risk in the whole analysis did not belong to the heaviest lifters. It belonged to the people who combined regular resistance training with substantial aerobic activity. Neither one alone matched the pair.

That result matters more to me than the two-hour figure, and it happens to describe the argument this site has been making for years. The federal Physical Activity Guidelines for Americans have said the same thing since 2018: 150 to 300 minutes a week of moderate aerobic activity plus muscle-strengthening work on two or more days. Not one or the other. Both, as a system.

I train low-intensity most days and lift on three to five of them. For a long time I framed the daily low-intensity work as the boring maintenance layer underneath the real training. This study reframed it for me. The aerobic base is not the supporting act. In the joint analysis, it is half of what produced the best result.

There was one oddity worth flagging honestly: cancer mortality did not follow the same pattern. Lower risk showed up only at the low resistance bands — one to 29 and 30 to 59 minutes a week — not across the range. I do not have a good explanation for that, and neither, as far as I can tell, does anyone else. When a subgroup behaves strangely in a large observational study, the responsible reading is usually "noise, or something unmeasured," not "the real signal." I mention it because leaving it out would make the picture tidier than it is.

The Floor Almost Nobody Talks About

We spent this entire article arguing about a ceiling. Almost nobody is anywhere near it.

According to the National Health Interview Survey, 31.9% of American adults did muscle-strengthening activity on two or more days a week in 2020. Among men aged 45 to 64 it was 29.9%; among women in that band, 23.8%. Both fall further after 65.

Read that next to the study and the practical message inverts. The debate over whether the answer is 60 minutes or 120 is a debate among the roughly quarter of middle-aged adults who already lift. For the other three-quarters, the distance from zero to anything is where nearly all of the available benefit sits — the steep part of the curve, before it starts to bend.

And the stakes rise with age. Adults lose somewhere in the range of 3 to 8% of muscle mass per decade after about 30, with the rate accelerating after 60. I want to be careful with that figure: it is a population estimate from the sarcopenia review literature, with enormous individual variation, not a constant that describes you. But the direction is not in dispute, and resistance training with adequate protein is the best-supported thing anyone has found to slow it.

Which is the Eat pillar showing up in a training article, as it always does. Loading the muscle is a request. Protein is whether the request can be honoured. Doing one without the other is asking your body to build something and not delivering the materials.

What I Changed, and What I Didn't

I did not cut anything. That surprised me, because I went into this expecting the study to give me permission to do less.

What changed was the accounting.

I had been treating my lifting volume as an open-ended good — more is more, and if the week is generous, add. Now I treat it the way I treat any resource that has stopped compounding: fund it to the level that does the job, and stop pretending the marginal hour is doing what the first hour did.

That reframing is not really about training. It is the same discipline I have spent twenty years trying to hold in platform engineering, and I am not consistently good at it there either.

Every leader I know has a project that stopped paying and kept getting funded. Not because anyone did the math and concluded it was worth it — because the effort had become identity. We had always staffed it. Cutting it would mean admitting the last two quarters of it bought less than the first two. So it stays, and the cost lands somewhere invisible: the initiative that never got started, the person who never got the time.

The two-hour ceiling is a small, physical version of the same lesson. When a curve flattens, the honest response is not to push harder on it. It is to ask what you would do with the hour if you stopped.

For me the answer was not to reclaim the hour. It was to stop mentally charging that hour to my health and start charging it where it belongs — to the fact that I like lifting, that a heavy set clears my head better than anything else I have found, and that a body that can still do hard things at 54 is worth having for reasons that have nothing to do with a hazard ratio.

That is a better reason than the one I had. It is also a more durable one. Reasons built on a specific number die when the number moves — and this number has already moved once, from 30-to-60 in 2022 to 90-to-120 in 2026. It will move again.

When a curve flattens, the honest response isn't to push harder on it. It's to ask what you'd do with the hour if you stopped.