"Let's take five" is usually followed by "we'll work better afterward." When I started looking into this, I assumed that was the link I would end up confirming.
What came through most clearly in the pooled experiments, though, was something else.
① "Short break" here means ten minutes or less
A 2022 meta-analysis pulled this literature together: 19 experimental and quasi-experimental studies from the past thirty years, yielding 22 independent samples and 2,335 participants.
Micro-breaks were defined as interruptions to work lasting no more than ten minutes. Getting a coffee, looking out a window, standing up to stretch.
Three things were measured: vigor, fatigue, and performance.
② The clearest change was that people were less worn out
Fatigue gave the cleanest result of the three. People who got a short break were less tired afterward.
The number in the paper is 0.35, which doesn't mean much on its own. Assuming the two groups' scores spread out in the usual bell shape, it translates like this:
▸ Pick one person who took a break and one who didn't, at random. The break-taker is the less tired of the two about 6 times out of 10.
▸ If the two conditions made no difference at all, that figure would be 5 out of 10.
Vigor came out almost identical, at 6 times out of 10. But the projected range for vigor stretches into "no effect" territory for future studies, so it isn't as solid as fatigue.
③ What stood out was the consistency more than the size
Those 22 samples came from eight countries. The participants were office workers, nurses, call-centre staff, and students.
What people did during the break varied just as much. Some studies had them move around, some gave them a different cognitive task, some had them look at nature, some simply had them relax. Break lengths ran from three minutes to ten.
Across all that variation, the fatigue results didn't scatter. Between-study variance was 0%, and the projected range suggests future studies would land in the same direction.
An effect size of 0.35 is conventionally filed under "small." But small is always relative to what you are paying, and here the price is five minutes.
The effect also barely depended on conditions. Break length didn't change it, nor did student versus employee, nor laboratory versus real office.
④ Performance was a different story
Across 15 samples, the performance effect was not statistically significant. On the same scale it works out to roughly 5.5 times out of 10, close enough to chance that you can't separate the two.
The tempting line here is "short breaks don't make you more productive." The next paragraph of the same paper makes that impossible to write.
When the authors checked for publication bias, the signal turned up only for performance. Studies finding no effect may have been published less often. Correcting for that actually pushed the estimate into statistical significance.
So what can be said about productivity is not "there is no effect" but "it doesn't show up as clearly as fatigue does."
⑤ It showed up least after mentally demanding work
There was one more split in the performance results, depending on what people had been doing right before the break.
▸ After cognitively demanding work — no detectable effect (9 samples)
▸ After creative work — a significant improvement (3 samples)
▸ After routine clerical work — the largest effect (only 2 samples, and the authors flag it as one to read cautiously)
Longer breaks also tended to produce larger performance effects. The authors concluded that recovering from cognitively depleting work may require breaks longer than ten minutes.
So expecting five minutes to restore your edge after a long stretch of hard thinking isn't an expectation this evidence supports. That is separate from what those five minutes do for fatigue.
⑥ So what should you do during the break?
The authors say plainly that they couldn't answer this. There were too few studies to break down by what people actually did, and they note this is the single most practically important question they failed to address.
For mood and fatigue, though, they say the answer is already visible: any activity that decouples you from the work seems to do it, whatever the type.
Which matches what we saw above. Moving, looking outside, or just zoning out, the fatigue results held.
⑦ If you're weighing whether to take one right now
▸ If you're skipping breaks because work will pile up — this evidence gives no support to the idea that pushing through wins on output. What it does show, consistently across eight countries and twenty-two samples, is that the people who broke were less worn out.
▸ If breaks don't seem to restore your focus — if what you were doing was cognitively heavy, that was never what five minutes could deliver. Shifting your expectation from output to fatigue fits the evidence better.
▸ If you're unsure what to do with the time — it barely mattered. Stepping away from the work is the part that counts.
The question gets simpler if you change it. Not "will a break make me more productive" but "how worn out am I right now." This evidence answers the first faintly and the second fairly clearly.
How often to break things up is something I covered separately.
→
barosit.com/community/p/d2f6b0c8-3e79-4a45-b81c-7d3e2f8a5b94
This is precisely where BaroSit sits. It tells you when you've held one position too long, and as this evidence shows, there is no prescribed answer for what to do with that time.
To be straight about it, the effect of on-screen reminders isn't large either. A 2025 analysis of 18 randomised trials with 1,164 office workers found computer prompts cut sitting time by about 12 minutes a workday, with no significant change in work, musculoskeletal, or metabolic measures.
→ Have a look at
barosit.com if you're curious.
The evidence on sitting time and movement is collected at
barosit.com/science.
Sources
• Albulescu, Macsinga, Rusu, Sulea, Bodnaru & Tulbure, 2022 · PLoS One 17(8):e0272460 — 22 samples from 19 experimental and quasi-experimental studies, 2,335 participants. Micro-breaks defined as no longer than ten minutes. Vigor 0.36 (95% CI 0.16–0.55) and fatigue 0.35 (0.19–0.50) were significant; performance 0.16 (−0.04–0.37) was not. Fatigue showed 0% between-study heterogeneity with a prediction interval of 0.16–0.53. For performance, Egger's test was significant and the bias-corrected estimate was 0.22 (0.02–0.43). By preceding task, performance was −0.09 (−0.39–0.20, 9 samples) after cognitive work, 0.38 (0.11–0.64, 3 samples) after creative work, and 0.56 (0.01–1.12, 2 samples) after clerical work. No moderator was significant for vigor or fatigue. The authors state they could not analyse what participants did during breaks. The "times out of 10" figures above are this effect size converted to a common-language effect size
• Leppe-Zamora, Ramos-Fuster, Muñoz-Monari, Roa-Alcaino & Sarmiento, 2025 · Int J Behav Nutr Phys Act 22(1):75 — 18 randomised trials, 1,164 office workers. Computer prompt software reduced workday sitting time by 12.46 minutes (95% CI −18.12 to −6.80) and increased steps by 1,030 per workday. Work-related, musculoskeletal, and cardiometabolic outcomes favoured prompts but were not statistically significant. Certainty of evidence low to moderate
The evidence here concerns short breaks during work, and both vigor and fatigue were self-reported by participants. All included studies are pre-pandemic literature.
This article is general information, not medical advice. If fatigue persists for weeks despite rest, or comes with symptoms such as weight change or shortness of breath, please see a clinician.