BACKGROUND--The increased demand of full polysomnographic studies, not only for diagnostic purposes but also for continuous positive airway pressure (CPAP) titration, has produced long waiting lists. Simpler methods are therefore needed to avoid having to refer all patients for full polysomnography. The hypothesis that CPAP therapy for the sleep apnoea/hypopnoea syndrome (SAHS) can be performed exclusively on the basis of recording night time respiratory variables was tested. METHODS--The level of CPAP in a group of 41 patients (three women) of mean (SD) age 52 (10) years, body mass index 31.5 (4.4) kg/m2, and apnoea/hypopnoea index (AHI) 53(16) events/hour was measured. During a two week period CPAP titration was performed in a random order in two settings: (1) in the sleep laboratory using full polysomnography; and (2) in the respiratory ward using equipment which continuously recorded and displayed pulse oximetry, airflow, chest and abdominal motion, and body position. The level of CPAP was increased progressively until apnoea, hypopnoea, snoring, and thoraco-abdominal paradox disappeared. RESULTS--No differences in CPAP levels (9.34(2.2) versus 9.68 (2.1) cm H2O) were found between full polysomnography and night time respiratory recordings. The accuracy of the measurement of both procedures showed good agreement. Only one patient showed a significant difference in CPAP level requirements between the two methods. CONCLUSIONS--Night time respiratory recording is sufficient to permit a reasonable choice of CPAP levels to abolish all the respiratory disturbances in most of the patients studied.