The purpose of this study was to determine factors increasing daytime PaCO2 or PaO2 in obstructive sleep apnoea syndrome patients (OSAS) with normal pulmonary function tests. Anthropometric, pulmonary function tests, arterial blood gases and sleep polygraphic data were analysed retrospectively in 218 OSAS patients (apnoea-hypopnoea index > 15 h(-1); 18 females, 55 +/- 11 years): 125 patients had abnormal pulmonary function tests, i.e. one or more flow or volume under 80% or above 120% of predictive value (group I) and 93 had normal pulmonary function tests (group II). Hypercapnia was defined as PaCO2 > or = 6.0 kPa and hypoxia as PaO2 < 9.3 kPa. Patients with abnormal pulmonary function tests were more hypoxic and hypercapnic, more obese, and had a higher apnoea-hypopnoea index (P<0.05). Seventeen patients of group I and four of group II were hypercapnic (13.6% and 4.3%, respectively). Thirty-one patients in group I (24.8%) had a PaO2 < 9.3 kPa and six (6.5%) in group II. Stepwise multiple regression analysis showed that in group II, only two factors were correlated with PaCO2: mean apnoea duration and FRC (respectively: c=0.228, P<0.001; c=0.006, P=0.108); and only two with PaO2: mean apnoea duration: (c=-0.218, P=0.029) and BMI (c=-3.72, P<0.0001). Daytime hypercapnia is present in 4.3% and daytime hypoxia in 6.5% of patients with occlusive sleep apnoea syndrome and normal pulmonary function tests. These alterations in blood gases in OSAS with normal pulmonary function tests should be considered as OSAS severity criteria.