Sleep hypoventilation is common in hypercapnic chronic obstructive pulmonary disease (COPD) and may contribute to daytime hypercapnia. The contributions of respiratory drive and respiratory mechanics to alterations in minute ventilation VI during sleep in this group have not been assessed. We assessed VI, breathing pattern, upper airway and total lung resistance (UAR, RL), intraoesophageal diaphragmatic EMG (EMGoes), intrinsic positive end-expiratory pressure (PEEPi), dynamic compliance (Cdyn), pressure-time product of oesophageal pressure (PTPoes), tension-time index of the diaphragm (TTIdi), end-expiratory lung volume (EELV) and respiratory drive (assessed as the slope of Poes versus time in the isovolumetric interval before PEEPi is overcome). Measurements were made in wakefulness and non-rapid eye movement (NREM) sleep, on 76%N2/24%O2 and on 76%He/24%O2 (heliox). Satisfactory data for analysis were obtained in 10 patients; five had measurements on heliox. VI fell from (mean (S.E.M.)) 8.84(0.46) to 6.64(0.91 l min(-1), P = 0.011) between wakefulness and stage II sleep, due to a fall in tidal volume. No changes were seen in PEEPi, Cdyn, EELV, PTPoes, TTIdi, EMGoes or respiratory drive. UAR increased (3.11(0.8) to 10.24(2.96) cm H2O l(-1) s (P=0.013) but RL was unchanged. UAR was reduced on heliox (5.20(1.67) to 3.45(1.35) cm H2O l(-1) s, P=0.049) but VI during sleep did not increase. PTPoes (350.2(51.0) to 259.4(46.3) cm H2O s min(-1), P=0.016), TTIdi (0.13(0.02) to 0.10(0.02) P=0.04), and respiratory drive (20.48(4.69) to 15.02(4.57) cm H2O s(-1), P=0.01) were all reduced. This suggests respiratory drive alters to maintain a preset VI in sleep, irrespective of load, at least while the fatigue threshold of respiratory muscles is not exceeded.