Sedation for fibreoptic bronchoscopy should produce optimal conditions for the operator, patient comfort and rapid recovery allowing early discharge home. We have compared a regimen producing 'light' sedation with a more traditional regimen producing 'deep' sedation. Seventy-six patients undergoing fibreoptic bronchoscopy under topical anaesthesia were randomized to receive either light sedation with the short acting opiate, alfentanil (median dose 1.1 mg, range 0.5-2.6 mg) or deep sedation with a combination of papaveretum (median dose 10 mg, range 5-15 mg) and diazepam (median dose 8 mg, range 0-20 mg). Both techniques gave equally good operating conditions, although patients given alfentanil coughed less than those given papaveretum and diazepam (U = 2.814 P less than 0.01). Patients recorded their degree of apprehension on a visual analogue scale prior to sedation and the actual degree of comfort experienced after recovery. There was no significant difference between apprehension or comfort between the groups. This was despite a higher degree of amnesia for an irrelevant object shown during the bronchoscopy in the deeply sedated group (chi 2 = 21.084 P less than 0.001). Patients given alfentanil performed significantly better in a modified Romberg test (chi 2 = 4.357 P less than 0.05) and a visualisation test (t = 3.035 P less than 0.01) two hours after the bronchoscopy. Alfentanil produced good operating conditions, patient comfort, less cough and a more rapid recovery, compared to the deep sedation regimen, and is an ideal sedative for fibreoptic bronchoscopy.