Why clenching a muscle helps you let it go, and what a new sleep review could not settle

The instruction sounds backwards the first time anyone hears it. To relax a muscle, squeeze it hard first.

That is the engine of progressive muscle relaxation, a routine the American physician Edmund Jacobson published in 1929 and which has been taught in sleep clinics, oncology wards and stress courses ever since. You work upward through the body in groups. Squeeze the calves for a few seconds, let go, wait, notice what changed. Then the thighs, the belly, the hands, the shoulders, the face. A review published on 6 August in Frontiers in Public Health gathered fourteen randomized trials of that routine and asked whether it helps adults sleep.

What the review found

Kehan Li and colleagues, working across three Malaysian universities, searched five databases through May 2026 and pooled 957 participants. The result was a benefit of 1.24 in Hedges g, a standard way of expressing effect size, with a 95 percent confidence interval spanning 0.77 to 1.71 points of improvement and no part of it crossing over into harm (Frontiers in Public Health, 2026). In the paper those three figures are printed as minus 1.24, minus 1.71 and minus 0.77, because the sleep questionnaires used in these trials score lower when sleep gets better. Every one of them points the same way.

By convention an effect of that size is large. It also came with a heterogeneity figure of 85.5 percent, which is the review's own way of saying the fourteen trials disagreed with each other a great deal, and that the pooled number is an average across studies that were not measuring quite the same thing in quite the same people. Both halves of the population studied improved, those with cancer and those without, with the larger effect in the group without.

The two numbers I keep looking at

The first is a dose test. The authors ran a meta regression asking whether people who did more sessions slept better than people who did fewer, and the answer came back at p equals 0.2106. No relationship. More was not better. For a method whose usual failure mode is a person quietly deciding they must not be doing enough of it, that is worth sitting with.

The second is the age split, and it is where a headline could go wrong. Among adults under 55 the benefit reached 1.40 and was clearly significant. Among adults aged 55 and over the benefit was 1.17, and it did not reach significance. Read fast, that looks like a technique that stops working. Read properly, it is not. The two point estimates are close neighbours; what differs is precision, because the older group's interval was so wide that it stretched from a large benefit all the way across zero and out the other side. That is the signature of too few trials, not of an absent effect. Absence of evidence, as the phrase goes, and the authors are careful about it in their own conclusion.

Why the squeeze is there at all

Here is the part that interests me as a practitioner rather than a reader of statistics. Jacobson's clench is not the medicine. It's a contrast device. Most of us cannot feel a moderate amount of muscular effort at all, because it has been running so long that it has become the background, the way you stop hearing a fridge; what we can feel is a large change, so the method manufactures one, deliberately, in order to make the release loud enough to register.

Two or three times a year someone arrives in my lessons saying they have done tense and release nightly for years and it has gone quiet on them. The routine still runs. Nothing much happens. What that usually tells me is that the contrast has been learned, and the person now needs the other approach to the same problem: instead of raising the effort until the difference is unmistakable, you lower everything until the room goes quiet and small differences become audible on their own. That's the direction the Feldenkrais Method® works in, and it is a genuinely different mechanism from Jacobson's, not a better one. Each has its strengths. People who find one of them dull often find the other one immediately interesting, which is information about them rather than about the methods.

Here is a short version of the second approach. It asks for six minutes, no equipment, and no squeezing whatsoever.

What to do with this

If tensing and releasing gets you to sleep, keep doing it. Nothing in this review says otherwise, and fourteen trials pointing the same way is not a small thing. If it used to work and no longer does, the meta regression offers a clue about which lever not to pull: adding sessions was not what separated the people who improved from the people who did not.

And if what keeps you awake is the sense of a body that will not put itself down, that pattern has its own long history in the research on chronic stress and muscle tension, and it responds to attention more reliably than it responds to force. The link between sleep disruption and physical tension runs in both directions, which is part of why a method aimed squarely at muscles ends up being scored on a sleep questionnaire at all. Persistent sleep trouble, snoring that worries the person next to you, daytime exhaustion that rest does not touch: those belong in a conversation with your physician, and gentle evening practice sits alongside that conversation rather than in place of it. In the Feldy online movement program, the bedtime ritual is built on exactly the logic above, done lying down, in the dark, at low effort.

Fourteen trials, 957 people, and the one group the review could not resolve is the group most likely to be lying awake at half past two reading about it. Somebody should run those trials.