Forty patients, scheduled for elective surgery in a thoracic and vascular surgical unit, were anaesthetized by a total intravenous anaesthesia technique using midazolam and fentanyl. Subsequent reversal of anaesthesia by the specific benzodiazepine antagonist flumazenil was evaluated in a double-blind trial. The patients were observed in the recovery room postoperatively until the next morning, and their recovery was repeatedly evaluated during the first 240 min after anaesthesia. Six patients in the placebo group required an oral airway or an endotracheal tube during the first hours of recovery, whereas none who received flumazenil did. The respiratory rate was significantly higher after flumazenil than placebo during the first 4 h postoperatively (Anova, P less than 0.01). Blood pressure and heart rate were not different between the two groups. The degree of sedation, orientation in time and space and the ability to cooperate were significantly superior after flumazenil than placebo (Anova, P less than 0.01). Some degree of resedation was observed in both groups, affecting 95% of the patients who received flumazenil against 30% after placebo (Mann-Whitney, P less than 0.05). No adverse reactions attributable to the use of flumazenil were encountered.