BACKGROUND There is evidence from animal studies suggesting the existence of a ceiling effect for buprenorphine-induced respiratory depression. To study whether an apparent ceiling effect exists for respiratory depression induced by buprenorphine, we compared the respiratory effects of buprenorphine and fentanyl in humans and rats. METHODS In healthy volunteers, the opioids were infused i.v. over 90 s and measurements of minute ventilation at a fixed end-tidal PCO2 of 7 kPa were obtained for 7 h. Buprenorphine doses were 0.7, 1.4, 4.3 and 8.6 microg kg(-1) (n=20 subjects) and fentanyl doses 1.1, 2.1, 2.9, 4.3 and 7.1 microg kg(-1) (n=21). Seven subjects received placebo. In rats, both opioids were infused i.v. over 20 min, and arterial PCO2 was measured 5, 10, 15 and 20 min after the start of fentanyl infusion and 30, 150, 270 and 390 min after the start of buprenorphine infusion. Doses tested were buprenorphine 0, 100, 300, 1000 and 3000 microg kg(-1) and fentanyl 0, 50, 68 and 90 microg kg(-1). RESULTS In humans, fentanyl produced a dose-dependent depression of minute ventilation with apnoea at doses > or = 2.9 microg kg(-1); buprenorphine caused depression of minute ventilation which levelled off at doses > or = 3.0 microg kg(-1) to about 50% of baseline. In rats, the relationship of arterial PCO2 and fentanyl dose was linear, with maximum respiratory depression at 20 min (maximum PaCO2 8.0 kPa). Irrespective of the time at which measurements were obtained, buprenorphine showed a non-linear effect on PaCO2, with a ceiling effect at doses > 1.4 microg kg(-1). The effect on PaCO2 was modest (maximum value measured, 5.5 kPa). CONCLUSIONS Our data confirm a ceiling effect of buprenorphine but not fentanyl with respect to respiratory depression.