Bronchoscopy was performed in 76 outpatients using local anaesthesia plus diazepam 0.2 mg kg-1 i.v. or midazolam 0.05 or 0.1 mg kg-1 i.v. Patient co-operation and ease of bronchoscopy were good in all patients. Two hours after injection, 67% (diazepam 0.2 mg kg-1), 36% (midazolam 0.05 mg kg-1) and 75% (midazolam 0.1 mg kg-1) of the patients failed to recall the insertion of the bronchoscope. Similarly, when asked on the following day, only 22%, 52% and 8%, respectively, of the patients remembered bronchoscopy. Two hours after sedation, the patients' performances in three psychomotor tests were similar to those measured before sedation in each group, but the patients' ability to stand steadily and walk along a straight line reverted to normal significantly (P less than 0.05) more slowly in patients receiving midazolam 0.1 mg kg-1 than in the patients given diazepam. The results suggest that midazolam offers no advantage over diazepam in terms of speed of recovery of psychomotor function, when doses of similar potency are given for bronchoscopy.