To prevent delayed emesis induced by cisplatin (mean dose 90 mg/m2), 120 consecutive patients were randomized to receive, in a 7-day crossover design, oral metoclopramide (20 mg q.i.d.), dexamethasone (1 mg q.i.d.) or placebo (two tablets q.i.d.) starting 24 hours after the end of chemotherapy. Complete protection from nausea, but not from vomiting. was significantly increased by both dexamethasone and metoclopramide with respect to placebo. Important prognostic factors favoring the appearance of delayed emesis were incomplete protection from vomiting during the first 24 hours after cisplatin, female gender, and highest cisplatin doses. Tolerability of both drugs was good. Larger and randomized controlled trials are necessary to identify better preventive treatment of delayed emesis induced by cisplatin.