In a double-blinded study, we have investigated psychomotor recovery after three sedation schemes during spinal anaesthesia for transurethral resection of prostate. Thirty-nine patients were allocated randomly to receive i.v. midazolam only, i.v. midazolam with antagonism by flumazenil or an infusion of propofol. The psychomotor tests were of long and short term memory, critical flicker fusion threshold, attention and concentration, cognition, mental arithmetic, and sedation and anxiety levels, and were performed before operation and at 30 min, 60 min, 2 h and 24 h after operation. Free recall of a picture learning task was the most sensitive test, showing a significant impairment in all groups, persisting for more than 2 h. Propofol was associated with the most rapid recovery and least impairment compared with preoperative baseline. Performance improved in all the tests after antagonism with flumazenil, but was still significantly impaired compared with baseline in tests of picture recall and letter deletion. Significant decreases in performance in all tests were detected 1 h after flumazenil. No increase in anxiety was recorded after flumazenil. We conclude that infusion of propofol is the technique with the most rapid recovery, while only brief and incomplete antagonism of the effects of midazolam may be expected with flumazenil.