Seventy-five breast cancer patients scheduled to receive a first course (in a new cycle) of cyclophosphamide, fluorouracil, and doxorubicin (FAC) or epirubicin (FEC) participated in a double-blind crossover study to compare the antiemetic efficacy and safety of ondansetron (GR38032), a 5-hydroxytryptamine3 (5-HT3) receptor antagonist, and metoclopramide. Ondansetron was given as an 8 mg loading dose (4 mg intravenously [IV] plus 4 mg orally) before chemotherapy followed by 8 mg every 8 hours orally for 3 to 5 days. Metoclopramide was given as an 80 mg loading dose (60 mg IV plus 20 mg orally) before chemotherapy followed by 20 mg every 8 hours orally for 3 to 5 days. A "period" interaction in the analysis of emetic response in the first 24 hours necessitated a parallel group analysis of first treatments only, 68 patients being assessable for this parameter. In the first 24 hours, complete or major control (zero to two emetic episodes) of emesis was achieved in 30 of 35 (86%) patients receiving ondansetron and in 14 of 33 (42%) patients receiving metoclopramide (P less than .001). Ondansetron was also more effective in reducing acute nausea. On days 2 to 3, the complete or major responses were significantly better with ondansetron (81% v 65%; P = .033), but there was no statistical difference in the control of nausea. There was a significant patient preference for ondansetron (63% v 26%; P = .001). Extrapyramidal reactions were observed in two metoclopramide treatments; both treatments were otherwise well tolerated. These results are consistent with serotonin (5-HT), being a significant neurotransmitter of cyclophosphamide/doxorubicin- or epirubicin/fluorouracil-induced emesis.