It is well known that obstructive sleep apnoea is especially frequent in the morbidly obese. In these subjects diurnal chronic hypercapnia, whose mechanism is still debated, may be present. Our study was performed to evaluate the prevalence and the mechanism of diurnal hypercapnia in the morbidly obese affected by obstructive sleep apnoea. From a population referred to our centre because of suspicion of sleep related breathing disorders, we selected 285 subjects without cardiopulmonary, neuromuscular or endocrinological diseases: 89 (36 M and 53 F, aged 46+/-13 years) had body mass index (BMI) > or = 40 kg m(-2) (MO group: morbidly obese subjects) and 196 (99 M and 97 F, aged 48+/-16 years) had BMI <40 kg m(-2) (NMO group: non-morbidly obese subjects). Then the MO group was divided into three subgroups: normocapnic subjects without obstructive sleep apnoea, normocapnic subjects with obstructive sleep apnoea, hypercapnic subjects with obstructive sleep apnoea; while we found no hypercapnic subject without obstructive sleep apnoea. All subjects underwent anthropometric evaluations and bioelectrical impedance analyses, respiratory function tests and arterial blood gas analysis, a modified version of the Sleep and Healthy questionnaire and a full night polysomnography. Our results showed that hypercapnia (PaCO2 > or = 45 mm Hg) associated with obstructive sleep apnoea [respiratory disturbance index (RDI) > or = 10 h(-1)] was found in 27% of the morbidly obese subjects, but only in 11% of the nonmorbidly obese ones (P<0.01). The comparison among the three subgroups, in which we divided the morbidly obese subjects, shows that those with hypercapnia and obstructive sleep apnoea had significantly more important ventilatory restrictive defects [forced vital capacity (FVC)% of pred 73.27+/-14 81 vs. 82.37+/-16.93 vs. 87.25+/-18.14 respectively; total lung capacity (TLC)% of pred 63.83+/-16.35 vs. 79.11+/-14.15 vs. 87.01+/-10.5], a significantly higher respiratory disturbance index (RDI 46.34+/-26.90 vs. 31.79+/-22.47 vs. 4.98+/-3.29) a longer total sleep time with oxyhaemoglobin saturation<90% [total sleeptime (TST)SaO2<90% 63.40+/-33.86 vs. 25.95+/-29.34 vs. 8.22+/-22.12] and a lower rapid eye movement (REM) stage (9.5+/-1.2 vs. 14.0+/-0.9 vs. 17.05+/-1.2) than normocapnic subjects with obstructive sleep apnoea or subjects without obstructive sleep apnoea. The best model to predict PaCO2 resulted from a combination of TSTSaO2<90% (r2 = 0.22, P<0.001), forced expiratory volume in 1 sec (FEV1)% of pred (r2 = 0.09, P<0.01), FVC % of pred (r2 = 0.075, P<0.01). In conclusion our study suggests that diurnal hypercapnia is frequently associated with obstructive sleep apnoea in the morbidly obese without chronic obstructive pulmonary disorder (COPD) and that ventilatory restriction and sleep related respiratory disturbances correlate to diurnal hypercapnia.