The purpose of this study was to compare the antiemetic efficacy of three 5-HT3 antagonists (granisetron, ondansetron, tropisetron) plus dexamethasone for the prevention of acute emesis induced by high-dose cisplatin chemotherapy. This was a randomized, open label, crossover study. Recruited into the study were 94 chemotherapy-naive patients of whom five were excluded because chemotherapy was not given, noncisplatin regimen was used instead, or presence of anticipatory vomiting. The remaining 89 evaluable patients were mostly (86.5%) male, and were all treated for head and neck cancers. The antiemetic regimens consisted of 1) granisetron 3 mg i.v. and dexamethasone 20 mg i.v. on day 1 (GRADEX); 2) tropisetron 5 mg i.v. and dexamethasone 20 mg i.v. on day 1 (TRODEX); and 3) ondansetron 8 mg i.v. and dexamethasone 20 mg i.v. to be followed by ondansetron 8 mg p.o. x 2 on day 1 (ONDEX). Patients were randomized to receive one of the three regimens in the first cycle, and treatment was crossed over to the other two regimens in subsequent cycles. Antiemetic efficacy was assessed using self-report diaries recording the number of vomiting episodes as well as duration and severity of nausea within the first 24 hours. Complete response was defined as no vomiting with or without mild nausea, and major response was defined as one vomiting episode and/or moderate to severe nausea. Major efficacy refers to either complete or major response. A total of 219 cycles was given to 89 patients: 16 received one cycle only, 16 received two cycles, and 57 received three cycles. No carryover effects were observed between cycles. Using pooled data from all cycles, the complete response rates to GRADEX, TRODEX, and ONDEX were 81%, 68%, and 71%, respectively (p = 0.11); the corresponding major efficacy rates were 91%, 93%, and 86%, respectively (p = 0.36). When only the first cycle was considered, the complete response rates to GRADEX, TRODEX, and ONDEX were 81%, 75%, and 74%, respectively (p = 0.58); the corresponding major efficacy rates were 92%, 94%, and 84%, respectively (p = 0.38). Analysis of the crossover data showed that the majority of patients achieved complete response or major efficacy with the different pairs of regimens, and there were no significant differences between different regimens in terms of complete response or major efficacy. The only exception was GRADEX versus TRODEX, in which 15.5% of patient achieved complete response with GRADEX as compared with 1.7% with TRODEX (p = 0.025). The majority of patients (53%) did not report any preference, whereas 14% preferred GRADEX, 15% preferred TRODEX, and 18% preferred ONDEX. The three 5-HT3 antagonists, when used in combination with steroids, had similar major efficacy for prophylaxis against cisplatin-induced acute emesis. Although GRADEX was superior to TRODEX in terms of complete response, this may not be of clinical significance. The choice of antiemetic regimens should therefore depend on patient preference and drug cost.