OBJECTIVE: To evaluate the validation of the AutoSet system in the diagnosis of sleep apnea syndrome(SAS). METHODS: One hundred and twenty patients were studied simultaneously performed both with the AutoSet and the polysomnography(PSG). The apnea index(AI) and apneas + hypopneas index (AHI) were compared between the two methods by Bland and Altmen method. RESULTS: A significant correlation was found between the AHI assessed by the AutoSet and by the PSG (r = 0.92, P < 0.001). The respiratory irregularity index (RII) of the AutoSet was well correlated with the arousal index (Ai) from the PSG (r = 0.72, P < 0.001). The sensitivity and specificity of the AutoSet in diagnosing sleep apnea were 94% and 80% for AHI > or = 5/hour, 92% and 85% for AHI > or = 10/hour, 91% and 92% for AHI > or = 15/hour, 91% and 91% for AHI > or = 20/hour. CONCLUSIONS: The AutoSet system is a sensitive and valuable tool for identifying patients with sleep apnea and can make a diagnosis of sleep apnea for these patients with AHI > or = 30/hour, but has some limits especially for mild to moderate sleep apnea.