Nasal obstruction is associated with abnormal breathing during sleep. To investigate this we measured ventilation and isocapnic hypoxic and rebreathing hypercapnic ventilatory responses in 9 awake normal men, with and without artificial nasal occlusion. Resting breathing frequency was lower (P less than 0.05) with mouth (12.5 +/- 1.0 [SEM]) than with nose (15.1 +/- 1.3 b/min) breathing, due to prolongation (P less than 0.05) of expiratory time with mouth breathing (mouth 3.25 +/- 0.35, nasal breathing 2.41 +/- 0.37 sec). Resting tidal volume was similar for both routes, thus minute ventilation was lower (P less than 0.01) mouth breathing (8.43 +/- 0.44) compared with nose breathing (9.37 +/- 0.47 L/min). Ventilatory responses were greater with mouth than nose breathing both for hypercapnia (mouth 2.29 +/- 0.21, nose 1.58 +/- 0.18 L/min/mm Hg CO2; P less than 0.01) and for hypoxia (mouth 1.08 +/-0.16, nose 0.91 +/- 0.21 L/min/% SaO2; P = 0.10). In 6 subjects measurements were repeated before and after upper airway lignocaine anaesthesia, which abolished the differences in respiratory timing and drive between the breathing routes. It is suggested that there may be upper airway flow receptors which influence respiratory timing.