A double-blind double-dummy, comparative study was carried out in 30 patients receiving highly emetogenic fast dose rate, single fraction total body irradiation prior to bone marrow transplantation. Patients were randomised into one of two groups, receiving either granisetron, a specific 5-HT3 antagonist or a combination of metoclopramide, dexamethasone phosphate and lorazepam to assess the comparative efficacy of the two regimens in the control of irradiation-induced nausea and vomiting. After 24 h eight patient (53%) treated with granisetron showed a complete response compared with two patients (13%) in the comparator group. The control of vomiting by granisetron, over both 24 h and 7-day periods (P = 0.001 and P = 0.004), was significantly better than that seen with the comparator and required significantly fewer rescue doses. The safety profiles in the two groups appeared similar, with the exception that granisetron produced less drowsiness than the comparator.