We carried out a randomized, single-blind, cross-over trial to compare the antiemetic effect, for both acute and delayed emesis, of granisetron plus dexamethasone (GRN+Dx) with that of high-dose metoclopramide plus dexamethasone (HDMP+Dx). Fifty-four patients with primary or metastatic lung cancer, given single-dose cisplatin (> 80 mg/m2) chemotherapy more than twice, were enrolled in this study. They were treated with both HDMP+Dx and GRN+Dx in two consecutive chemotherapy courses. On day 1, patients experienced a mean of 2.5 (SD = 4.3) and 0.1 (SD = 0.4) episodes of vomiting in the HDMP+Dx and the GRN+Dx groups, respectively (P = 0.0008). Complete response rate on day 1 was 45 and 90% in the HDMP+Dx and the GRN+Dx groups, respectively (P = 0.0001). Patients treated with GRN+Dx had a tendency to suffer more episodes of vomiting than the HDMP+Dx group on days 2-5, but it was not statistically significant. Twenty-four patients (57%) preferred the GRN+Dx treatment and 14 patients (33%), HDMP+Dx. In the HDMP+Dx group, nine patients (21%) had an extrapyramidal reaction, and 5 patients (12%) had constipation that lasted for at least two days. In contrast, no patients had extrapyramidal reactions, and 18 patients (43%) had constipation in the GRN+Dx group (P < 0.01). GRN+Dx was more effective than HDMP+Dx only in preventing the acute emesis induced by cisplatin. An effective treatment for delayed emesis is still needed.