A total of 1442 patients who had major gynaecological surgery were recruited into three multi-centre studies using a standard general anaesthetic technique in order to assess the efficacy of various doses of orally administered ondansetron in the prevention of postoperative nausea and vomiting. A total of 1257 patients were included in this analysis: 420 received oral formulations of placebo and 212, 296 and 329 received ondansetron 1, 8 and 16 mg, respectively. The following factors were measured in these studies and were considered to have a possible influence on the proportion of patients experiencing postoperative nausea and vomiting: age of patient; volatile anaesthetic; intraoperative dose of fentanyl; postoperative dose of morphine; country; anaesthesia duration; neuromuscular blocker; neuromuscular block antagonist; pre-medicant; recovery time , type of surgery; antiemetic treatment; body weight. Using a process of elimination based on logistic regression techniques, the factors found to be the most important influences on the frequencies of nausea and vomiting were antiemetic treatment, type of surgery, neuromuscular blocker, country, volatile anaesthetic and age. A statistically significant interaction between type of surgery and age was observed. Adjusted probabilities of nausea and vomiting were obtained from the model, including all the above factors, together with the type of surgery by age interaction. Ondansetron 8 mg showed the smallest adjusted probability of nausea (0.54) and vomiting (0.34) and placebo the greatest (nausea 0.75, vomiting 0.61). A similar pattern of adjusted probabilities over neuromuscular blocking agents was seen for nausea and vomiting, with the greatest occurring in patients receiving pancuronium (nausea 0.74, vomiting 0.57) and the least in patients receiving alcuronium (nausea 0.59, vomiting 0.38). For country effects, the United Kingdom exhibited the greatest adjusted probability of nausea (0.75) and vomiting (0.54) and Germany the smallest (nausea 0.49, vomiting 0.34). Adjusted probabilities of nausea and vomiting were similar for enflurane and isoflurane. With increasing age, the adjusted probabilities of postoperative nausea and vomiting appeared to increase for patients having abdominal hysterectomy and laparotomy. This pattern was not repeated in patients having vaginal hysterectomy. (Br. J. Anaesth. 1993; 71: 517–522)