The best exercise for neck pain may matter less than how often you actually do it

If you have gone looking for the right exercise for neck pain, you have probably met a wall of confident, contradictory advice. A new analysis pooling 51 trials suggests the confident part is where the trouble starts.

Published in PeerJ in late July, the review gathered 51 randomized controlled trials covering 4,140 people with nonspecific neck pain, the ordinary kind with no clear structural cause, and compared exercise approaches against each other rather than testing them one at a time (Guo et al., PeerJ, 2026). Just under three quarters of the participants were women.

Four approaches were associated with less neck related disability than control conditions. Mind body exercise, which in this analysis meant yoga and Pilates, showed a reduction of 0.363 on a standardized scale. Resistance training came in at 0.359. Flexibility training at 0.352. Coordination training at 0.328. Those are four different numbers, and the temptation is to read them as a leaderboard. They are not one. The uncertainty ranges around each estimate overlap almost entirely, which in plain terms means this analysis does not establish a winner among the four.

It is worth being equally clear about the size of what was found. A standardized difference of roughly 0.35 is small to moderate. This is not a result saying that exercise makes neck pain go away. It says that several quite different ways of moving are associated with modestly less disability, and that none of them separated from the others.

The more interesting signal sits in the dose, not the type. The authors' exploratory analysis suggested reductions appearing from around 100 MET minutes per session, with larger study level effects around 300 to 400, and a similar pattern by frequency: something showing up at roughly two sessions a week, more at five or six. They flag all of this cautiously, because study level dose data is heterogeneous and easy to read too much into. Still, the shape of it is suggestive. How much and how often may be carrying more of the signal here than which.

The ranking question is the wrong question

This is the part I find clarifying as a Feldenkrais Method® practitioner, because the question I am asked most often is a ranking question. Is yoga better than Pilates for my neck. Should I be doing tai chi instead. Is the Alexander Technique the one that actually gets at this.

Read against this analysis, that framing does not hold up well. Yoga, Pilates, tai chi, the Alexander Technique, and the Awareness Through Movement® lessons I teach are all methods you practice on your own, each with a genuinely different mechanism and its own strengths. Yoga and Pilates in these trials combined postures, motor control, breathing, and interoceptive attention in one package, which is roughly why the reviewers grouped them as mind body work. Resistance training loads tissue. Flexibility work changes tolerance to range. Coordination training rehearses control. These are not the same thing wearing different names. But on this outcome, in this population, they landed within touching distance of one another.

What I see in clients points at why that might be. The people who get somewhere with a long standing neck are rarely the ones who found the correct method. They are the ones who found something they could come back to on a Tuesday evening, when nothing hurts badly enough to be motivating and nothing yet feels like it is working. Neck and shoulder tension that has been around for years is not usually waiting for the right exercise to be identified. It is waiting for enough repetitions of something tolerable.

That is where the dose finding meets the human problem. A method you find punishing will not reach two sessions a week, whatever its mechanism. The honest case for gentle work is not that its mechanism outranks the others. It is that gentleness is what makes the frequency reachable, and frequency is where this analysis found its clearest pattern.

What to do with this

Less pressure on the choice, and more attention to whether what you choose is something you will still be doing in three months. If you already have a practice you like, nothing here gives you a reason to trade it for a better one. If you have been stalled trying to pick correctly, pick the one you would not dread.

For desk related neck pain in particular, the practical read is smaller and more often, rather than heroic and rare. Ten quiet minutes most days is closer to what the dose curve rewards than one demanding session on a Sunday.

Two cautions belong here. If your neck pain arrived with an injury, with numbness or weakness in an arm, or with anything a clinician has been following, this is not a decision to make alone. Whatever you practice sits alongside physiotherapy and clinical care, not in place of it. And whichever method you land on, staying inside the range that feels easy, pleasant, and comfortable is what keeps it repeatable.

None of this makes the choice unimportant. It makes it personal rather than technical. The variable worth attending to is not which of four roughly equivalent approaches you selected. It is whether you are still doing it some months from now. On this evidence, working gently enough to keep going is not a lesser version of the work. It may be the version that lasts long enough to count.