Ventilatory effects of propofol, used as a sole agent for the induction and maintenance of general anaesthesia, were studied in 14 healthy unpremedicated patients. Subarachnoid anaesthesia was established before induction of general anaesthesia. Induction was with propofol 2.5 mg kg-1 given while the patients breathed 100% oxygen. We intended to start an infusion of propofol 100 micrograms kg-1 min-1; maintain it for at least 25 min; make a first set of quasi-steady-state observations; double the infusion; and repeat observations after 25 min. The single induction bolus plus single rate infusion was not totally satisfactory: further boluses were usually needed. At induction there was apnoea in all but three patients, sometimes lasting more than 3 min; hyperventilation before induction, combined with hyperoxia, probably exaggerated this. Established ventilatory rates were generally 30% higher than awake. One patient became bradypnoeic. Tidal volume and minute ventilation, and the Tl:Ttot ratio, were reduced. Doubling the infusion rate had no clear effect on frequency or tidal volume, but it further reduced the Tl:Ttot ratio and caused an increase in PE'CO2 of 1 kPa. The ventilatory response to carbon dioxide was 58% of baseline awake control (95% confidence limits +/- 26%) at the lower infusion rate, with further slight depression when the infusion rate was doubled. Doubling the rate of infusion of propofol did not give twice the effect on ventilation, and probably is not giving twice the "depth" of anaesthesia. We cannot say if this is for pharmacokinetic or pharmacodynamic reasons.