Do Mouth and Throat Exercises Treat Sleep Apnea?
A Cochrane review of nine trials found a large reduction in apnoea severity and a clear improvement in sleepiness. The certainty ratings attached to those two findings are different, and that is the whole story.
Exercises for the tongue, soft palate and pharyngeal muscles — myofunctional or oropharyngeal therapy — are one of the few non-device approaches to sleep apnea with a Cochrane review behind them. That makes this unusually assessable.
What the review covered
Nine randomised trials, analysing 347 participants in total, of whom 69 were women and 13 were children (Cochrane Database of Systematic Reviews, 2020).
Adult mean ages ranged from 46 to 51, daytime sleepiness scores from 8 to 14, and severity from mild to severe apnea. Study duration ranged from two to four months, and nine further studies were ongoing [Cochrane 2020].
What it found, with the certainty attached
The certainty grades are the point here, and they are usually stripped out when this evidence gets quoted.
Daytime sleepiness — probably reduced. Against sham therapy, Epworth Sleepiness Scale scores fell by a mean difference of −4.52 points (95% CI, −6.67 to −2.36) across two studies and 82 participants, rated moderate-certainty evidence [Cochrane 2020].
That is the strongest finding, and a 4.5-point drop on the Epworth is clinically meaningful — the scale runs 0 to 24 and the threshold for excessive sleepiness is usually taken as 10 or 11.
Apnoea severity — may be reduced, and the estimate is large. A mean difference of −13.20 events per hour (95% CI, −18.48 to −7.93), again from two studies and 82 participants — but rated low-certainty evidence [Cochrane 2020].
Thirteen events an hour is a big number. It would move many people down a severity class. It is also low certainty from 82 participants, which means the true effect could be considerably smaller.
Sleep quality — may improve, mean difference −3.90 points (95% CI, −6.31 to −1.49), from a single study of 31 participants, low certainty [Cochrane 2020].
Snoring — may have little to no effect, and the evidence is very uncertain, with a standardised mean difference of −0.53 [Cochrane 2020].
The snoring finding deserves emphasis
Because most people arrive at these exercises looking for less snoring, and that is the outcome the review is least able to support.
Sleepiness has moderate certainty. Apnoea severity has low certainty with a large estimate. Snoring frequency has "little to no effect" with very uncertain evidence. Those are three different conclusions and they should not be collapsed into "exercises work."
If your goal is a quieter bedroom, this is weaker ground than if your goal is feeling less sleepy. Throat exercises for snoring.
What the review could not assess
Two gaps, both stated plainly by the authors:
None of the studies assessed accidents, cardiovascular disease or mortality [Cochrane 2020]. So there is nothing here about whether myofunctional therapy affects the outcomes that make apnea worth treating in the first place.
No included study reported adverse events — the reviewers looked and found none reported [Cochrane 2020]. That is an absence of data rather than a demonstration of safety, though the intervention is exercises, so the prior risk is genuinely low.
Add the duration: two to four months. Nothing here speaks to whether the effect persists, or whether people keep doing the exercises once a study ends. Adherence to a daily exercise regimen over years is a real question that a four-month trial cannot answer.
How it compares
Against CPAP, this is not a substitute. CPAP overrides airway collapse mechanically at any severity, with far larger and better-established effects. A −13.2 AHI reduction at low certainty is not equivalent to what pressure does. CPAP versus an oral appliance.
As an adjunct, it is more interesting. Myofunctional therapy targets upper airway muscle function — one of the four physiological traits that determine how apnea behaves in a given person. Someone whose dominant problem is poor dilator muscle responsiveness is, mechanistically, the person it should help most. Nobody selects patients that way yet. Why treatments work for some people and not others.
Against doing nothing, it has moderate-certainty evidence for reduced sleepiness, no cost, and no known harm. For mild apnea, or alongside other treatment, that is a reasonable proposition.
Tier 2 overall — real randomised evidence, small samples, mixed certainty, and the outcomes that matter most unmeasured. How we grade evidence.
What this means practically
Do not replace established treatment with exercises. Particularly not with moderate-to-severe apnea, and not on the strength of a low-certainty AHI estimate.
It is a defensible addition, especially if you have mild apnea, cannot tolerate CPAP, or are waiting for an appointment. CPAP alternatives.
Expect months, not weeks. The trials ran two to four months. Anything promising results in days is not describing this evidence.
Get a proper programme rather than a video. The trials used structured protocols delivered with instruction, typically by a speech-language pathologist or trained therapist. Whether unsupervised imitation produces the same effect is untested.
And re-test if you rely on it. If exercises are your treatment for diagnosed apnea, the only way to know whether your AHI actually moved is a repeat sleep study — the same logic that applies to an oral appliance, which reports nothing about itself. Oral appliance titration · How to read your sleep study.
The line
Exercises are not a consumer product and this page is not a recommendation to skip assessment. No over-the-counter product treats obstructive sleep apnea, and neither does an exercise routine you have not had verified. Snoring versus sleep apnea.