Objective/background In obstructive sleep apnea (OSA), long-term adherence to treatment is crucial. This prospective single-center study investigated factors associated with long-term adherence to mandibular repositioning device (MRD) therapy. Patients/methods All OSA patients who had MRD treatment initiated in the previous year were prospectively contacted to evaluate long-term effectiveness and compliance. Long-term adherence was based on continuation of treatment (yes/no). Predictors of long-term adherence were analyzed using an adjusted multivariate analysis. Results Median follow-up was 1002 days (interquartile range: 668–1345) in 279 patients (age 58 [50–64] years); 63% of patients were continuing MRD treatment with a good efficacy, tolerability and compliance over time. In some patients, relapse of nocturia was observed while efficacy was maintained for snoring and somnolence. In adjusted multivariate analysis, significant predictors of continuing MRD treatment were early ≥50% reduction in AHI (odds ratio [OR] 2.73, 95% confidence interval [CI] 1.466–5.10; p = 0.0002) and complete symptom resolution (OR 3.83, 95% CI 1.74–8.48; p = 0.0014). In the 37% of patients who stopped MRD treatment, median treatment duration was 351 (174–752) days. The main reasons for late stopping of treatment were inefficacy (26.2%), discomfort (25.2%) and side effects (21.4%). Conclusions After three years, MRD was effective for the two-thirds of OSA patients who continued treatment. Relapse of nocturia might be an early signal of MRD wear that may explain long-term cessation of treatment in some patients. Short-term control of OSA by MAD was predictive of long-term efficiency. The major criteria were a ≥50% reduction in AHI and complete symptom resolution at short-term evaluation.