This study was undertaken to determine the minimum effective dose of granisetron, 5-hydroxytryptamine type 3 receptor antagonist, for the prevention of post-operative vomiting in children undergoing general inhalational anaesthesia for surgery (inguinal hernia and phimosis). In a randomized, double-blind manner, 120 children, ASA physical status I, aged 4–10 years, were assigned to receive placebo (saline) or granisetron at three different doses (20 μg kg−1, 40 μg kg−1, 100 μg kg−1) intravenously immediately after inhalation induction of anaesthesia (n=30 of each). A complete response, defined as no emesis and no need for another rescue antiemetic during the first 24 h after anaesthesia, occurred in 57% with placebo, 67% with granisetron 20 μg kg−1, 90% with granisetron 40 μg kg−1 and 90% with granisetron 100 μg kg−1 respectively (P<0.05; overall Fisher's exact probability test). No clinically important adverse events were observed in any of the groups. Our results suggest that granisetron 40 μg kg−1 is the minimum effective dose for the prevention of emesis after paediatric surgery, and that increasing its dose to 100 μg kg−1 provides no demonstrable benefit.