Eighty-seven per cent of surgical patients receiving undiluted diazepam experienced pain on injection while 6-16%, depending on the dose, manifested evidence of clinical thrombophlebitis. This was improved when diazepam, 10 mg, was diluted to 20-40 ml with intravenous solution. In contrast, lorazepam appeared to have minimal irritative or injurious effects on veins whether undiluted or diluted. In view of these results and clinical studies reporting a higher patient acceptance of lorazepam than diazepam, lorazepam may be a superior drug for use in anaesthesia.