Nocturnal oximetry can show nocturnal oxygen desaturation. This examination was proposed as an investigation for the early detection of the sleep apnoea syndrome (SAS). We have compared the results of nocturnal oximetry and polysomnography in 329 consecutive patients seen in the department of thoracic medicine for the early detection of the SAS between June 1990 and June 1995. The diagnosis of SAS was confirmed at the time of polysomnography using an hypopnoea/apnoea index (IAH) greater or equal to 15 per hour. Two parameters of oximetry were well correlated with IAH less than 15 per hour: if the mean oxygen saturation is greater than 92% and for less than five per cent of the time of the examination there was a saturation of less than 90%. The sensitivity was 89.7% and the specificity was 57.8%. Among the 48 false positive cases on oximetry 17 patients were found to be suffering from COPD and 31 patients were probably suffering from a syndrome of upper airways resistance or possibly from the hypoventilation obesity syndrome. Amongst the 22 false, negatives to oximetry 10 non COPD patients with an IAH of greater than 30 per hour and diurnal somnolence had important anomalies of the oro-pharyngeal pathway as the origin of their nocturnal apnoea. The 12 other false negatives were patients with moderate SAS with an IAH of between 15 and 20 per hour. Logistical analysis has shown the association of the two oximetric criteria (mean oxygen saturation or percentage of time with a saturation of less than 5%) with clinical criteria (body mass index and formation on diurnal somnolence from a questionnaire) would enable a probable diagnosis of SAS in 75% of cases. Our study shows that nocturnal oximetry used an early diagnosis test, associated with clinical and respiratory function data enables the number of requests for polysomnography to be reduced.