The efficacies of granisetron plus dexamethasone and granisetron alone in controlling nausea and vomiting during two consecutive cycles of moderately emetogenic chemotherapy given for up to 5 days were compared in a two-centre, randomised, double-blind, placebo-controlled crossover study. In all, 110 evaluable patients received either dexamethasone, 20 mg i.v., or matching placebo, plus open-label granisetron, 3 mg i.v., given on each chemotherapy day. At cycle 2, patients crossed over to the alternative treatment; 72 patients completed the crossover. In these 72 patients, the complete response rates over 24 h for granisetron plus dexamethasone and granisetron plus placebo in cycle 1 were 87% and 70% (ns), respectively. In cycle 2 the complete response rates over 24 h were 73% and 62% (ns). Combining the two cycles, the complete response rates over 24 h were 80.6% (granisetron plus dexamethasone) and 65.3% (granisetron plus placebo; P = 0.015). Granisetron plus dexamethasone was significantly more effective in terms of times to less than complete response (P = 0.041), to first episode of moderate/severe nausea (P = 0.04), to first episode of vomiting (0.03) and to use of rescue medication (P = 0.02). Adverse events tended to be minor, with asthenia and insomnia the most common. Of those patients who expressed a preference, 67% preferred granisetron plus dexamethasone (P < 0.05). A single dose of dexamethasone added to granisetron thus enhances the efficacy of granisetron alone in preventing nausea and vomiting after moderately emetogenic chemotherapy.