We have studied 64 ASA I and II patients (aged 20-60 yr) to determine if nitrous oxide affects sevoflurane requirement for achieving 50% probability of no movement in response to verbal commands (MACawake). Patients were allocated randomly to one of four nitrous oxide concentration groups (0, 20, 40 and 60 vol.%). Patients in each group received sevoflurane at two different end-tidal concentrations according to a predetermined randomization table. After steady state sevoflurane and nitrous oxide concentrations had been maintained for at least 15 min, patients were assessed as being awake or asleep. The MACawake for sevoflurane was 0.63% and this was reduced significantly in a non-linear manner by increasing nitrous oxide concentration. A 50% reduction in MACawake was produced by a nitrous oxide concentration of 45%. The reduction in MACawake by nitrous oxide was non-linear; the interaction coefficient between nitrous oxide and sevoflurane being significantly less than zero (P = 0.0238), indicating that the reduction in MACawake by nitrous oxide was smaller than would be expected from simple additivity and that nitrous oxide antagonized the effects of sevoflurane in preventing response to verbal commands.