To study the need for and mode of naloxone reversal after liberal use of fentanyl during anaesthesia, a fixed‐dose schedule of fentanyl (mean 6.5 μg/kg/h) was applied intraoperatively to 60 patients, randomly divided into four treatment groups. After surgery and reversal of muscle relaxation, end‐tidal CO2 was allowed to rise to 8% by manually assisted ventilation. Thereafter, 0.04 mg increments of naloxone were given i.v. every 3 min until spontaneous respiration maintained ETco2 below 8% (group I) or below 6.5% (group II). After similar intravenous reversal to that in groups I or II, an additional intramuscular dose of naloxone, identical to that given i.v., was given to groups III and IV after awakening from nitrous oxide.If ETco2 remained above 6.5%, mean spontaneous minute ventilation was lower than before anaesthesia and significantly (P<0.05) less than if ETco2 had decreased below that value. This difference had disappeared 45 min postnal‐oxone. Arterialized venous Pco2 showed normocapnia (6.0 kPa) in patients with ETco2 <6.5%, hut slight hypercapnia (6.6 kPa) in patients with ETco2>6.5% after intravenous naloxone. After 15 min, the patients with additional intramuscular naloxone had significantly (P<0.05) lower values of Pco2 than those with intravenous naloxone only.Postoperatively, a positive correlation (R = 0.35) between the total dose of naloxone and pain intensity could be demonstrated (P<0.01). As acute postoperative pain could not be avoided after naloxone reversal to normoventilation, titration of naloxone after liberal use of analgesics is recommended with the help of capnography, until an acceptable ETco2 of about 7% is maintained.