The Autoset autotitration system (Resmed, Australia) was designed as an aid for determination of the effective level of CPAP. Two studies have shown that the 95% pressure centile during a night with Autoset, after excluding any periods with severe leak (P95), was well correlated with the pressure determined manually according to the reference method. However, the suitability of P95 under Autoset for treatment with conventional CPAP needed to be evaluated outside of the context of titration. The objective of this study was to measure the compliance and efficacy of CPAP at P95 determined by Autoset, after 3 months. Twenty-four patients who had undergone routine autotitration by Autoset under polysomnographic control in the laboratory for previously untreated OSAS were evaluated after 3 months of treatment with CPAP at P95. This evaluation was based on compliance, daily usage, determined by reading the hourly counter, Epworth sleepiness score and home respiratory recording under CPAP. Polysomnography under Autoset showed a significant reduction of abnormal respiratory events (apnea-hypopnea index (AHI): 5.7 (+/-4.6) vs 69.6 (+/-29.8) n/h; p<0.001). P95 was between 8.5 and 15 cm H2O (mean: 11.2 (+/-1.6)). Two patients were lost to follow-up. At 3 months, 18 out of 22 (82%) patients declared that they were still using their machine with a daily usage of 5.25 (+/-1.82) h. The Epworth sleepiness scale was significantly decreased (5 (+/-4.1) vs 11.4 (+/-5.4); p<0.05). P95 recommended by Autoset remained effective on abnormal respiratory events with a low AHI (4.1 (+/-3.2) n/h) and a low 4% oxygen desaturation index (3.6 (+/-4.5) n/h) under CPAP. This study confirms the efficacy of autotitration and the suitability of P95 determined by Autoset for long-term treatment by conventional CPAP at home.