BackgroundTo explore the feasibility, the efficacy, and the mechanism of mandibular advancement devices (MAD) in the treatment of persistent sleep apnea after surgery.MethodsNineteen patients who failed uvulopalatopharyngoplasty (UPPP) or UPPP plus genioglossus advancement and hyoid myotomy (GAHM) were given a non-adjustable MAD for treatment. All patients had polysomnography (PSG) at least 6 months post-UPPP with and without the MAD. Seventeen patients had computed tomography (CT) examinations.ResultsAfter the application of MAD, the apnea hypopnea index (AHI) decreased significantly from 41.2 ± 13.1/h to 10.1 ± 5.6/h in the responder group. The response rate was 57.9 % (11/19). During sleep apnea/hypopnea acquired from sedated sleep, the cross-sectional area and anterior-posterior and lateral diameters of the velopharynx enlarged significantly from 4.2 ± 6.0 mm2 to 17.5 ± 15.3 mm2, 1.9 ± 2.3 mm to 6.5 ± 4.1 mm, and 1.1 ± 1.3 mm to 2.6 ± 2.1 mm, respectively (P < 0.01) in the responder group with MAD. The velopharyngeal collapsibility also decreased significantly from 83.3 ± 21.8 % to 46.5 ± 27.1 %. The glossopharyngeal collapsibility decreased from 39.8 ± 39.1 % to −22.9 ± 73.2 % (P < 0.05).ConclusionMAD can be an effective alternative treatment for patients with moderate and severe OSAHS after surgery. The principal mechanisms underlying the effect of MAD are expansion of the lateral diameter of the velopharynx, the enlargement of the velopharyngeal area, the reduction of velopharyngeal and glossopharyngeal collapsibility, and the stabilization of the upper airway.