What Causes Delayed Onset Muscle Soreness: Novelty, Not Effort

What soreness is telling you, and what it isn’t

You added Romanian deadlifts on Tuesday. By Thursday you cannot sit down properly. Meanwhile the squats you have run for two years, at a weight that would have wrecked you in your first month, produce nothing at all.

That contrast is the whole answer to what causes delayed onset muscle soreness, and it is not the answer most gym advice gives. Soreness does not track how hard you worked. It tracks how unfamiliar the work was.

A 2023 systematic review and meta-analysis in Sports Medicine makes that case about as directly as evidence can. It pooled studies where people did a deliberately easy bout of an exercise, chosen to sit far below what normally causes damage, and then did the hard version days later. The easy bout removed most of the soreness from the hard one. If effort were the cause, that could not happen.

What causes delayed onset muscle soreness
Why new exercises make you sore and familiar ones don’t
The timing is two to four days, not the morning of
Soreness is a bad gauge of anything
What it means if you already train
The limits of this evidence
What causes delayed onset muscle soreness: the practical answer
Common Questions About Muscle Soreness
How Pick It Up Helps
Further Reading & References

What causes delayed onset muscle soreness

Start with what the soreness is part of. Researchers group it under exercise-induced muscle damage: the set of things that go wrong in a muscle after unaccustomed hard work. Boyd and colleagues, whose review this article is built on, list the symptoms as localised soreness and swelling, a rise in creatine kinase (a protein that leaks into the blood when muscle cells are damaged), a drop in how much force you can produce, and movement simply feeling harder than it should. It sets in within about 48 hours and can last two to ten days.

The mechanical cause, in the review’s framing of the field, is lengthening under load. Eccentric contractions are the lowering half of a rep, where the muscle produces force while being stretched, and they do the most damage, because the muscle is being pulled apart while it is trying to hold on. That is why Romanian deadlifts, walking lunges and slow negatives are the exercises people remember two days later, and why a session of machine presses at the same effort usually isn’t.

What the review does not do is settle the fine mechanism. Several explanations compete for why a prior bout protects you: the connective tissue around the fibres realigning to transmit force better, cytoskeletal proteins upregulating to hold the sarcomeres together, antioxidant enzymes rising while inflammatory markers fall. The authors say plainly that the mechanisms “are still debatable, and require more exploration beyond the scope of this review.” So this is not a piece that will tell you which molecule makes your hamstrings hurt. Nobody can tell you that yet.

What is settled enough to act on is the input: soreness comes from loading a muscle in a way it is not currently used to, with the lengthening portion doing most of it. The strongest evidence for that framing is what happens when you remove the novelty.

Why new exercises make you sore and familiar ones don’t

Do a damaging bout of exercise, and the same bout done again weeks later hurts much less. That is the repeated bout effect, and it is long-established enough that Boyd’s introduction treats it as background rather than as a finding: it puts the protective window at two to eight weeks, with roughly six months before a muscle returns to its pre-exposure state.

The obvious problem with earning that protection is that you have to be wrecked once to get it. So a line of research asked whether the first bout could be made deliberately easy — a few three-second isometric holds, a set of eccentrics at 10% of maximum effort, five minutes of walking downhill — and still buy the protection. In the studies Boyd’s team cite, those interventions produced no measurable rise in creatine kinase and no soreness of their own.

This review is the first to pool that literature: 5,803 abstracts screened, 23 studies kept, 19 with data usable for the meta-analysis. Everyone in them was untrained, which matters a great deal and which the limitations section below comes back to.

The soreness result is not subtle. Across the studies measuring it, people who had done the light bout were substantially less sore at 24, 48 and 72 hours than people who went in cold, and still less sore at 96 hours. Every one of those gaps clears the threshold the authors themselves call “large” by a wide margin. The same pattern held for the other three things they measured: less creatine kinase in the blood, more of their maximum force retained, and more range of motion kept at the joint.

The meta-analysis forest plot for muscle soreness 48 hours after a damaging bout of exercise. Forty-seven comparisons are listed, most sitting left of the centre line, meaning the group that did a light pre-conditioning bout first was less sore. The pooled result at the bottom is minus 2.52, with heterogeneity of 89 percent

Muscle soreness 48 hours after the hard bout, the point where soreness usually peaks. Each row is one comparison between a pre-conditioned group and a control group; a row left of the centre line means the pre-conditioned group was less sore. The diamond at the bottom is the pooled answer. Note how far apart the rows are: that spread is the 89% heterogeneity on the line beneath, and it is why the authors rate their own certainty as very low. From Supplementary File 6 of Boyd et al. (2023), used under CC BY 4.0.

Run the causal logic backwards and it is hard to escape. The pre-conditioning bout was chosen specifically to be far easier than the bout it was protecting against — in several of the studies it registered on no damage marker at all — and it still took most of the soreness out of what followed. Whatever soreness is measuring, it is not the difficulty of the session that produced it.

The timing is two to four days, not the morning of

This is the part that changes what you do, and the part most people get backwards. Pre-conditioning done immediately before the hard work did essentially nothing. Done 2 to 4 days before, it produced the largest protection in the review. Done 1 to 3 weeks before, it still helped, but less. A warm-up set is not pre-conditioning; a separate light session earlier in the week is.

Four other patterns came out of the individual studies, and they all point the same way. The more the light bout resembles the hard one, the more it protects:

Eccentric beats isometric. Actually lowering the weight protects better than holding still against it.
More volume beats less. The review’s summary is 10 to 30 contractions rather than 2 to 6.
Higher intensity beats lower. Eccentrics at 80% of maximum protected better than at 40 or 60%, and 20% beat 10%. Downhill walking beat flat walking.
Longer muscle length beats shorter. Contractions with the muscle stretched out protected better than the same contractions in a shortened position.

There is a cost to that scaling, and the authors say so: the more aggressive versions of the light bout did produce some soreness and some creatine kinase of their own. It was still far below what the unprepared group went through after the hard bout.

How long the protection lasts depends on what you did to earn it. Between two and ten maximal isometric holds bought only 2 to 4 days. Higher-volume isometric work lasted several weeks. Eccentric work ran to weeks or months. And ordinary heavy strength training, through the standard repeated bout effect, may protect for months, which is the real reason your squats stopped making you sore.

Soreness is a bad gauge of anything

The temptation, once you know soreness is about novelty, is to flip it into a training rule: chase novelty, get sore, grow. That does not follow, and there is a cleaner reason to distrust soreness as a signal.

In a 2002 study, Nosaka and colleagues had 110 men perform 12, 24 or 60 maximal eccentric contractions of the elbow flexors — five times as many at the top end as at the bottom. The damage markers responded as you would expect: the bigger doses produced larger changes and slower recovery. Soreness mostly did not. On two of the three ways they measured it, the men who did 60 contractions were not reliably sorer than the men who did 12. One soreness measure did not correlate with any of the damage indicators at all, and the other two correlated only weakly. Their conclusion was that soreness is “a poor reflector” of eccentric exercise-induced muscle damage. (That paper is paywalled; we worked from its abstract, which states those results directly.)

Put the two findings side by side. Soreness can be removed without removing the work, and soreness barely moves when the work is multiplied by five. In both directions, the thing you feel two days later is a poor instrument for measuring what you actually did.

Neither paper says anything about whether being sore helps you grow; that question sits outside both of them. What they do establish is that soreness is not the readout you want. Sets and load are measurable, and they are what the weekly volume evidence connects to growth. Soreness is neither.

What it means if you already train

Three things to take from this, and the first is reassurance rather than a change.

If you are barely getting sore, nothing is wrong. You have been pre-conditioning for years without calling it that. The protection from regular heavy training is the strongest and longest-lasting version of this effect in the literature. An absence of soreness in a program you have run for months is what adaptation looks like from the inside.

When you introduce a genuinely new movement, front-load a light version of it. This is the practical takeaway and it costs almost nothing. If a new lengthened-position exercise goes into next week’s program, put two or three easy sets of it into a session 2 to 4 days beforehand — light load, full range, controlled lowering. Then run it properly. On this evidence you should arrive at the real session with most of the soreness already spent, and the sessions after it intact.

Protect the sessions that follow, not the soreness itself. Boyd’s review opens by noting that muscle damage costs you force and power output in subsequent sessions, degrades proprioception, and reduces how hard people are willing to train. That is the real bill for a wrecked Tuesday: a mediocre Thursday. If soreness has been dictating your schedule, the training frequency evidence is worth reading alongside this, because the weekly total is what does the work, so a session you have to move is a logistical problem rather than a lost week.

One thing not to do: treat soreness as a reason to keep rotating exercises. Rotating movements to feel something is how you spend every week paying the novelty tax and never collecting the protection. If progress has stalled, our guide to breaking through a plateau covers the levers that actually move it.

The limits of this evidence

This is where the paper needs handling with care, because its own verdict on itself is unflattering.

The authors grade their certainty as very low. Almost every outcome came out at the bottom rating of the standard framework for this, downgraded three ways: fewer than 800 participants per outcome, suspected publication bias in every outcome, and study-to-study disagreement above their threshold. That is not a stray caveat in a discussion section. It is the authors’ formal summary of how much weight their own numbers can bear. The individual studies were fine, rated good or excellent across the board on their appraisal checklist. It is the body of evidence that is thin.

Nobody in it lifts. Every participant was untrained, most with no resistance training for at least six months and many with none for a year. They were also almost entirely young men. And the review quotes a prior paper making the exact objection our lifters should make: pre-conditioning “has not yet been explored in a trained population and may not be effective, owing to the long-term prophylactic effects of eccentric contractions.” In other words, the reason it may not work for you is that you already have the protection. That is a limit on the tactic rather than on the explanation, but it is a real limit. Take the advice above as a cheap thing to try, not as something proven for you.

The exercises were mostly not exercises. Much of this literature runs on single-joint machine protocols — elbow flexors on a dynamometer, knee extensors at a fixed speed — plus downhill walking. The authors flag it themselves: it “remains unclear to what extent” the effect transfers to a real session of compound lifts. The most encouraging data point is one included study where an isometric leg press protected against a later bout of downhill running, which suggests the protection follows the muscle rather than the exercise.

The studies disagree with each other a lot. Heterogeneity sat at 89% for soreness at every time point, which is very high, and some comparisons in the forest plot above show almost no effect at all. Pooling protocols this varied into one number produces an average over quite different things. Where original data was not available, the team also digitised values out of published charts, which adds error they acknowledge.

So place it carefully, because the two halves of this paper do not carry the same weight. On the causal story — unaccustomed loading causes soreness, and prior exposure removes it — this study is restating a settled position rather than establishing one; the repeated bout effect is background material in its own introduction, and has been studied since long before this review. On the specific prescription of a deliberately light bout two to four days out, it is the first pooled attempt at the question, it says so, and it rates its own certainty as very low. Treat the first as reliable and the second as a promising thing to test on yourself.

What causes delayed onset muscle soreness: the practical answer

Novelty, biased toward lengthening under load. Soreness comes from loading a muscle in a way it is not currently accustomed to, and eccentric work causes the most of it. Not from effort: five times the eccentric work did not reliably produce more soreness.
Expect it within 48 hours, and expect it gone in 2 to 10 days. That is the window this review reports. Anything much longer or much sharper is worth a different conversation than a training article.
Prime a new movement 2 to 4 days ahead. Two or three easy sets, full range, controlled lowering. Immediately before does nothing; a week or more before does less.
Make the light bout resemble the hard one. Lowering rather than holding, more reps rather than fewer, a real load rather than a token one, and the stretched position rather than the shortened one.
Do not read soreness as a scoreboard. It does not scale with the work done, and it can be removed without removing the work. Track sets, load and reps instead.
Judge the intervention by the next session, not by the ache. Less damage meant more force and more range of motion retained across the following four days, which is the part that shows up in your training.

Common Questions About Muscle Soreness

What causes delayed onset muscle soreness? Loading a muscle in a way it is not used to, with the lowering portion of the movement doing most of it. The clearest evidence is that a deliberately light first bout, far below what normally causes damage, removes most of the soreness from a hard bout days later. The precise biological mechanism is still contested, and this review says so.

How long does DOMS last? This review puts exercise-induced muscle damage, soreness included, at an onset within about 48 hours and a duration of two to ten days.

Does being sore mean the workout worked? Nothing here supports that. Soreness barely changed when one study multiplied the eccentric workload fivefold, and it can be almost eliminated while the work stays the same. It is measuring unfamiliarity, not productivity.

Why am I not sore anymore? Because you are trained. Regular heavy lifting produces the strongest and longest-lasting version of the protective effect in this literature, and the review notes it may last months. The soreness stopping is the adaptation arriving.

Can I train a sore muscle? This paper does not test that. What it does report is that muscle damage reduces force and power output, impairs proprioception and lowers training effort in the sessions that follow — so the case for arranging your week to avoid it is about protecting the quality of those sessions.

Does a warm-up prevent soreness? Not as pre-conditioning, on this evidence. Light work done immediately before the damaging bout produced essentially no protection; the effect needed a day at minimum and worked best at 2 to 4 days.

Will priming a new movement work for me if I already lift? Unknown, and worth saying plainly. Every participant in this review was untrained, and the authors quote a prior review suggesting the tactic may add little for trained lifters precisely because regular eccentric work already protects them.

How Pick It Up Helps

The awkward part of this research is the scheduling. “Two or three light sets of the new movement, 2 to 4 days before you run it properly” is easy to understand and annoying to place. It means knowing what is coming next week, which session it lands in, and what “light” means in kilograms for a movement you have never done.

That is programming, and it is what Pick It Up does. You tell it your goal and the equipment you can reach, and it selects the exercises and builds monthly blocks, so a new movement enters your program on a schedule rather than on a whim. Every set arrives with a target weight, target reps and a target difficulty on a simplified RPE scale, which is how “light” becomes a number you can load. And because the targets update from the weight, reps and difficulty you actually report, a week where a new exercise leaves you beaten up is a week your subsequent targets know about.

Start a free trial and have all the science taken care of for you, so you can get back to building a better you.