Mandibular Advancement Device (MAD): What the Evidence Actually Shows

A mandibular advancement device (MAD), also called a mandibular repositioning appliance (MRA), is a dental-anchored oral appliance worn during sleep that holds the lower jaw in a forward position. Because the base of the tongue is attached to the lower jaw, advancing the jaw also pulls the tongue base forward, keeping the airway more open during sleep. A MAD requires enough remaining teeth to anchor to, and because it holds the jaw in an advanced position all night, it loads the temporomandibular joint (TMJ) in a way a device with no dental anchorage does not.

MAD devices come in two broad categories: custom-fitted devices, made from a dental impression and adjusted by a dentist over time, and boil-and-bite or over-the-counter (OTC) devices, molded by the user at home without individual fitting. As covered below, these two categories do not perform equivalently.

Sleep Apnea Mouth Guard and Mouth Guard for Sleep Apnea: Same Device?

“Sleep apnea mouth guard” and “mouth guard for sleep apnea” are common everyday search terms for the device described in this article. The clinical literature almost always uses “mandibular advancement device,” “mandibular repositioning appliance,” or “oral appliance” instead; it's the same device.

Oral Appliance for Sleep Apnea

“Oral appliance for sleep apnea” is the term used most consistently in the clinical guideline literature, including the current U.S. guideline discussed later in this article. It's the same MAD/MRA device category.

Mouthpiece for Sleep Apnea

“Mouthpiece for sleep apnea” is another common lay search term for the same device category the clinical literature calls a mandibular advancement device or oral appliance.

Yes, with real and well-documented reductions in AHI in randomized trials — but with an important real-world caveat that trial results alone don't capture.

A landmark 2013 randomized crossover trial in the American Journal of Respiratory and Critical Care Medicinedirectly compared MAD against CPAP in 108 patients who completed both arms, from a baseline AHI of about 25.6 events/hour. CPAP reduced AHI further on average (down to 4.5) than the MAD (down to 11.1, p<0.01). CPAP is the more powerful therapy on paper. But the same trial found patients wore the MAD significantly more hours per night than CPAP (6.5 vs 5.2, p<0.00001), the MAD outperformed CPAP on four SF-36 quality-of-life domains, and the MAD was non-inferior to CPAP on 24-hour blood pressure.

The more important number for a 2026 reader comes from a much larger, more recent dataset. A 2026 retrospective study in Sleep Medicine, following 595 real-world patients treated with oral appliance therapy at the Mayo Clinic, found only 10–28% of patients (depending on baseline severity) achieved an objective response — at least 50% AHI reduction with post-treatment AHI under 15 — and only about 20% got post-treatment AHI under 10. That's a much lower bar-clearing rate than the small, selected trial populations behind most efficacy numbers elsewhere in the MAD literature. The same cohort reported much higher subjective satisfaction than the objective numbers suggest (89% reported no or mild residual snoring, 90% reported improved sleep quality), illustrating why patient-reported improvement and an objectively normalized AHI aren't the same thing, and why follow-up sleep testing (below) matters. Predictors of better objective response: higher baseline AHI, female sex, and mandibular retrusion; other craniofacial features were not predictive.

One smaller finding: a single randomized trial of MAD in upper airway resistance syndrome (UARS) rather than full OSA — 30 patients, reported across two papers in 2017 and 2020 — found significant improvements in objective breathing measures (arousal-related events and flow-limitation time all decreased over 1.5 years) and in subjective stress scores, with no significant difference on cognitive testing. A single small trial in a narrower population, not a general OSA finding.

The 2013 crossover trial above remains the clearest single head-to-head comparison: CPAP reduces AHI more, but the MAD is worn more hours per night and can match CPAP on several quality-of-life measures. A more recent trial adds a cardiovascular data point. The CRESCENT trial, published in 2024 in the Journal of the American College of Cardiology, randomized 220 patients with OSA and hypertension to MAD or CPAP for six months and found MAD non-inferior to CPAP for reducing blood pressure. Two caveats: the non-inferiority was asymmetric — 24-hour mean arterial pressure fell significantly on the MAD (down 2.5 mmHg, p=0.003) but not significantly on CPAP in this trial, so “MAD is as good as CPAP for blood pressure” overstates what was shown. And one of the trial's senior authors holds a ResMed-funded endowed chair and a financial interest in SomnoMed, a MAD manufacturer, plus other device-company consultancies, which is worth disclosing.

Neither trial changes the underlying picture: CPAP remains the more powerful therapy for reducing AHI itself, and the U.S. clinical guideline (discussed below) does not position MAD as a general first-line alternative to CPAP. For readers deciding between staying on CPAP or trying an oral appliance, understanding your own CPAP data — pressure needs, residual AHI, usage hours — is a useful starting point before that conversation with a sleep physician; AirwayLab's CPAP data analysis platform and its guide on what counts as a good AHI on CPAP cover that side of the picture in more depth.

Because a MAD is anchored to the teeth, dental status is the central candidacy question. A 2002 study in the American Journal of Respiratory and Critical Care Medicine, which had two blinded maxillofacial surgeons examine 100 consecutive polysomnography-diagnosed patients, found an absolute dental contraindication to MAD therapy — almost entirely due to insufficient remaining teeth — in 34% of patients, with a further 16% needing close dental or TMJ supervision if treated. That figure needs context: it comes from a 2002 French cohort, and population dentition has generally improved and edentulism rates have generally fallen since then, so 34% is very likely an overestimate for a 2026 population. For patients without sufficient teeth for a dental-anchored device, one commonly cited alternative is a tongue retaining device, which needs no dental anchorage — though, as AirwayLab's companion article covers, that indication rests on clinical reasoning rather than dedicated trial evidence. Readers unsure how a MAD differs from a TRD, eXciteOSA-style tongue stimulation, or myofunctional therapy can find the full four-way breakdown in that companion article's disambiguation section.

Beyond dental status, two predictive approaches have been studied for identifying likely responders. A 2013 prospective study (the MATRx system) used a single night of remotely controlled jaw positioning to predict treatment outcome with 83–94% accuracy in a small, selected tertiary-clinic trial population of 67 patients. Whether that accuracy holds outside such selected populations is unknown: in the large, unselected real-world cohort discussed above, objective response was only 10–28%. That study's authors are affiliated with the company that commercializes MATRx, so it's best understood as a manufacturer-conducted, prospective, blinded validation study awaiting independent replication, not independently confirmed evidence. Separately, a 2014 Australian cohort study found patients who need a lower CPAP pressure tend to respond better to a MAD (9.7 vs 11.7 cmH2O, p<0.01). But a pressure cutoff derived from a Japanese population failed to transfer reliably to that Australian cohort, while a locally derived cutoff reliably ruled out response in that same population. CPAP pressure is a genuine predictive signal, but no cutoff should be treated as universal.

This is the clearest single head-to-head comparison in the MAD evidence base, and the basis for the guideline's custom-device preference. A 2008 randomized crossover trial in the American Journal of Respiratory and Critical Care Medicine directly compared a custom-fitted device against a thermoplastic (boil-and-bite) device in 35 patients who completed both arms. Treatment success in this 35-patient trial population was 60% with the custom device versus 31% with thermoplastic (p=0.02), a considerably higher figure than the real-world objective response rates discussed above; the thermoplastic device failed entirely in 69% of patients, mostly from not staying retained overnight; and 82% preferred the custom device (p<0.0001). Quoted directly, the authors' conclusion was that the thermoplastic device “cannot be recommended... nor used as a screening tool.”

Long-term MAD use produces measurable dental and occlusal changes, and two credible sources characterize how concerning those changes are quite differently — both deserve to be stated, not merged into one reassuring line.

A 2014 study in the Journal of Clinical Sleep Medicinefollowed 77 long-term MAD users for an average of 11.1 years and found statistically significant reductions in overbite (2.3mm average), overjet (1.9mm average), and mandibular crowding (1.3mm average), along with increases in intercanine and intermolar width, plus more anterior crossbite and posterior open bite. In the authors' words, “rather than reaching a discernible end-point, the dental side effects of MAS therapy continue with ongoing MAS use” (MAS = mandibular advancement splint, the review's term for a MAD): clinically significant and progressive, without a plateau. A separate 2014 narrative review by Sutherland et al. (a different paper from the Sutherland-led CPAP-pressure-predictor cohort study cited earlier in this article) reached a notably more reassuring conclusion on the same question, stating verbatim that “long-term dental changes do occur, but these are for the most part subclinical and do not preclude continued use.” Both are legitimate, differently framed readings of the same evidence; readers should know both perspectives exist rather than only the reassuring one.

Adherence Over Time

A 2025 study in the European Journal of Orthodontics, using objective, sensor-verified adherence tracking in 55 patients, found 80.0% adherent at one month, declining to 67.3% at six months. The abstract does not report a 12-month adherence percentage at all, and 38% of the original cohort (21 of 55) had dropped out entirely before 12 months, so a claim that adherence “remains substantial at 12 months” isn't supported; the honest summary is a decline through six months and a substantial dropout rate by one year, with no 12-month figure available.

The current U.S. clinical practice guideline on oral appliance therapy, published jointly by the American Academy of Sleep Medicine and the American Academy of Dental Sleep Medicine in 2015, makes six graded recommendations. Precision matters here, because MAD is sometimes described online as endorsed on equal footing with CPAP, which it is not. Only two of the six carry the guideline's highest (“standard”) certainty rating: prescribing an oral appliance rather than no therapy for primary snoring without OSA, and considering an oral appliance rather than no treatment for patients who are CPAP-intolerant or who prefer an alternative therapy. The other four — preferring a custom, titratable device over a non-custom one (consistent with the 2008 trial above), dentist oversight for occlusal and dental side effects, follow-up sleep testing, and periodic follow-up with both a dentist and a sleep physician — carry the lower (“guideline”) strength rating. This remains the current U.S. standard-of-care document as of September 2026; no more recent U.S. guideline supersedes it.

That follow-up-testing recommendation isn't a formality. A 2009 randomized study in Sleep and Breathingcompared self-guided device adjustment against objective, sleep-study-guided adjustment in 28 patients: at six weeks, 71% had either resolved (AHI under 5) or meaningfully improved (a greater than 50% AHI reduction with AHI still above 5), but 29% had neither improved nor resolved. A confirmatory sleep study detects whether a device is working; it does not itself normalize AHI, and neither self-report nor titration method reliably tells you it's working without that follow-up test.

One older evidence source is worth citing with its age disclosed rather than as current best evidence: a 2006 Cochrane systematic review of 16 randomized trials (745 participants) found oral appliances reduced AHI compared with a control device and improved sleepiness scores, while also confirming oral appliances were less effective than CPAP at reducing AHI. That review has never been updated, and it predates the modern era of custom, titratable devices and current CPAP comparators; its numbers are still cited, but always with 2006 attached, not presented as reflecting today's device generation.

Specific dollar figures and insurance coverage details vary too widely by device type, dental practice, region, and policy to state a fixed number here honestly. What the evidence does support: custom-fitted devices are the type shown in trials to meaningfully outperform boil-and-bite devices (see above), and they involve a dental evaluation and fitting rather than an over-the-counter purchase. If cost or coverage is a deciding factor, check directly with your dentist's office and your insurer.