Inflammation 02: Delayed-onset muscle soreness

Short answer. Soreness tells you something happened, not how much you will grow: it responds to novelty, contraction type, unfamiliar range and the muscle trained, none of which scale with the growth stimulus. Whether muscle damage contributes to growth is still open. Do not program for soreness, do not treat its absence as failure, and judge a program on whether load and reps move over months.

What it is

The ache that arrives a day after training and peaks around forty-eight hours. It is not the burn during a set and it is not the fatigue immediately afterward. It is a distinct, delayed sensitivity, worst on stretch and on pressure, and it fades over three to four days without doing anything.

Most people use it as a gauge: a sore session was a productive one. That reading is the subject of this piece.

The proposed mechanism

The first account was lactic acid, and it is wrong: lactate clears within about an hour, long before soreness arrives. The account that replaced it is microscopic damage to muscle fibers, followed by inflammation that sensitizes nerve endings in and around the tissue.

The second account survives, with a qualification that matters. The damage side of it appears to be connective tissue and structures around the fiber at least as much as the contractile machinery itself, and the sensation is a nociceptive one rather than a direct readout of structural injury. Inflammation is part of this, and current reviews treat inflammation as integral to adaptation, not as an obstacle to it, provided it is tightly regulated (Peake et al., 2017).

None of that tells you whether soreness tracks the size of the damage, and none of it tells you whether damage builds muscle. Those are the two questions people answer without noticing they have answered them.

What has actually been tested

The most useful body of work here comes from a series of trials that tracked untrained people through a full training program rather than a single bout, measuring cross-sectional area, muscle protein synthesis and damage markers at the same time.

Early size gains are not what they look like. In the first weeks, increases in cross-sectional area are largely damage-induced swelling rather than new contractile tissue (Damas et al., 2015). This is a measurement problem before a physiological one: ultrasound and magnetic resonance imaging (MRI) cannot distinguish a bigger muscle from a swollen one.

The time course was mapped in sessions, not weeks. Increases in cross-sectional area through roughly the first four sessions are mostly swelling; modest hypertrophy follows after around ten sessions; and only after around eighteen is the change dominated by true hypertrophy (Damas et al., 2018).

The synthesis response reorders itself. Myofibrillar protein synthesis measured after the earliest sessions did not correlate with the hypertrophy those participants eventually achieved. Synthesis measured later, once damage had subsided, did (Damas et al., 2016). The authors read this as the early response being directed at repair rather than growth.

Growth happens without the damaging part. Thirteen untrained men trained one arm with concentric-only curls and shoulder presses twice a week for four weeks. Small but detectable hypertrophy appeared within seven sessions, in the absence of eccentric muscle damage (Stock et al., 2017). There was no damaging comparison group, so it does not show that damage adds nothing. It shows damage is not required for early gains to start.

The conclusion the authors draw is that muscle damage is not the process that mediates or potentiates hypertrophy. That is a strong claim, and it has been contested. We come back to that below.

Separately, soreness does not track damage size. Severe soreness can occur with minimal markers of damage and vice versa. Muscle groups differ markedly in how sore they get from comparable work, and all of them grow.

Our reading

Soreness is a real sensation reporting on real events. It is not a dial, and it is not a scoreboard.

Two claims get conflated and only one survives. That soreness indicates something happened is uncontroversial. That the amount of soreness indicates how much happened, or how much you will grow, does not follow and is not supported. Soreness responds to novelty, contraction type, unfamiliar range, and which muscle you trained, none of which scale with growth stimulus.

Whether damage contributes to growth is open. The authors who argued damage is not the mechanism were challenged on interpretation, and they conceded that whether fiber-level disturbance is required in early training remains unresolved. A confident answer in either direction goes past the evidence.

What follows practically is narrow but useful. Do not program for soreness, do not treat its absence as failure, and do not read the first month of measurements as growth in either direction, because a meaningful share of early size change is fluid. Judge a program on whether load and reps are moving over months, not on how you feel two days after a session.

Where the goals diverge

If you are new to training, the early scale reading and the early tape measurement are both partly water. This is not a reason for discouragement. It is a reason not to build a conclusion about your program on the first month of measurements, in either direction. (→ Loading fundamentals 05: Progressive overload)

If you use soreness as a session-quality gauge, the gauge is measuring something, but not what you want it to measure. It responds to novelty, to eccentric emphasis, to unfamiliar range of motion, and to which muscle you trained. It does not scale with the size of the growth stimulus.

If you are avoiding soreness deliberately, you are not necessarily giving up progress. That does not make soreness worthless, and it does not license the reverse conclusion that the least uncomfortable program is the best one.

If you are training around a competitive schedule, soreness matters for a different reason, because it degrades the sets that follow. (→ Loading fundamentals 02: Reps in reserve) It degrades next-session quality regardless of what it signifies about growth, which makes managing it a scheduling problem rather than a physiological one.

Interventions that suppress soreness are covered separately. (→ Inflammation 01: Cold water immersion; Cold and heat exposure sub-hub; Inflammation sub-hub)

The case against

This is the most actively contested claim in the branch, and the dispute is on the record.

Schoenfeld and Contreras wrote a letter to the editor arguing that the interpretation and extrapolation in the review above do not support the inference drawn from it, and that the question remains open. The journal did not publish it; they released it as a preprint, which has not been peer reviewed (Schoenfeld and Contreras, 2018). Their central point is a design one: it is close to impossible to isolate muscle damage as an experimental variable, because any manipulation that changes damage also changes load, range, contraction type or volume. A study that was not designed to test whether damage contributes to growth cannot conclude that it does not.

They also flag a statistical move worth naming, because we made a version of it ourselves in an earlier draft. The review cited a meta-analysis comparing eccentric-only against concentric-only training as evidence of no hypertrophic difference. The comparison did not reach significance (p = 0.07, just above the conventional 0.05 threshold), and its confidence interval, the range of effects the data are compatible with, ran from -0.03 to 0.52, centered well above zero. That supports "not established," not "no difference," and reading it as the latter is a mistake in the same family as the one this article is about.

The authors of the review replied in a blog post, also not peer reviewed. They conceded a point that belongs in any fair summary: whether disturbances inside the muscle fiber and the repair and remodeling that follow are required in early training phases remains unresolved (Damas et al., 2018, reply).

Most of this evidence comes from one research group. The session-count model, the protein synthesis time course and the edema finding all originate with the same group and its collaborators. That does not make the work wrong, and the group has answered its critics publicly. It does mean the independent replication that would move a finding to settled has not happened.

An older exchange in the European Journal of Applied Physiology cuts at the measurement itself (DeFreitas et al., 2015; Damas et al., 2016b). It argues that the inability to separate swelling from true hypertrophy limits interpretation on both sides of the argument, not only for those claiming early growth.

What is not known

Whether mild damage is a precondition for later growth is, in the authors' own words, unresolved. There is no experimental design available that isolates damage without changing something else, which means this may stay unresolved for a while.

What soreness physically is remains open. A minority hypothesis places the origin in microdamage to sensory nerve endings rather than to muscle tissue. We mention it because it exists, not because we find it persuasive.

And there is no established threshold. Nobody can tell you how much soreness is too much, or at what point damage stops being incidental and starts interfering with the training that follows.

The practical warning

Soreness that persists beyond about a week, comes with severe swelling, or is accompanied by dark urine is not delayed-onset muscle soreness, and nothing in this article applies to it. That combination needs same-day medical attention rather than a recovery strategy. Rhabdomyolysis is the word to use when you get there.