PURPOSE To evaluate antiemetic efficacy and tolerability of granisetron, dexamethasone, and their combination over repeated courses of moderately emetogenic chemotherapy, and the influence of the prognostic factors on occurrence of nausea and vomiting. PATIENTS AND METHODS Four hundred twenty-eight consecutive cancer patients were entered onto a multicenter, randomized, double-blind study to compare granisetron 3 mg intravenously, dexamethasone 8 mg intravenously, and 4 mg orally every 6 hours for four doses, or the combination of dexamethasone plus granisetron at the same doses, administered for three consecutive cycles. Occurrence of nausea, retching, and vomiting was monitored for 24 hours after chemotherapy administration by a diary card. RESULTS Three hundred ninety-eight patients were assessable for clinical efficacy at the first cycle, 354 were assessable at the second cycle, and 322 were assessable at the third cycle of chemotherapy. Dexamethasone plus granisetron induced significantly greater complete protection from vomiting, nausea, and both nausea and vomiting than granisetron alone in all three cycles. With respect to dexamethasone alone, complete protection from vomiting was significantly greater at the first and second cycle, and complete protection from nausea and from both nausea and vomiting only at the first cycle. Complete protection did not differ significantly among the three cycles in patients receiving dexamethasone plus granisetron or dexamethasone alone, whereas it decreased significantly, at least for vomiting, in patients receiving granisetron alone. Protection obtained in the previous cycle of chemotherapy was the most important prognostic factor in the occurrence of nausea and vomiting. CONCLUSION The combination of dexamethasone plus granisetron offers the best antiemetic protection because of its greater efficacy with respect to the other two regimens at first cycle, and because its activity is maintained in the subsequent cycles of chemotherapy.