Cisapride is a substituted benzamide with gastrointestinal prokinetic effects presumed to be due to the enhancement of the physiological release of acetylcholine at the myenteric plexus. In a multicentre study, 189 patients with nonulcer dyspepsia (NUD) received single-blind placebo treatment for two weeks. A total of 123 patients with no or minimal response to placebo and epigastric pain of at least moderate severity and frequency were randomly assigned to one of the three parallel double-blind treatments for six weeks: cisapride 10 mg tid, cisapride 20 mg tid or placebo. The severity and frequency of individual symptoms (epigastric pain, heartburn, nausea, vomiting anorexia, postprandial discomfort, regurgitation, lower abdominal pain, bloating and constipation) were assessed on a four- and five-point categorical scale, respectively, by the investigator at three on treatment visits and by patients in a daily diary. Analysis of investigator and patient assessments for differences in symptom severity x frequency composite scores among the three treatment groups showed no statistically significant differences for individual symptoms or symptom clusters. As assessed by the investigator, and compared with baseline, cisapride 20 mg tid significantly (P < 0.05) improved epigastric pain, bloating and early satiety as well as improved the total symptom cluster. Investigator evaluation of the five most severe and frequent symptoms for each patient showed statistically significant improvement in each treatment group. For patient diary assessments, statistically significant within-treatment improvement of the total symptom cluster, the five most severe symptoms cluster, bloating and early satiety was observed for both cisapride 20 mg and placebo, whereas epigastric pain significantly (P < 0.05) improved in all three treatment groups. Investigator evaluation of global response (good+excellent) rate at the end of the six week treatment period was 38% for cisapride 20 mg, 47% for cisapride 10 mg and 33% for placebo. No statistically significant difference in this parameter among treatments was noted. Cisapride was well tolerated at both doses with a side effect profile comparable with that of placebo. It is concluded that in this double-blind multicentre study with a single-blind two-week placebo run in phase, cisapride 10 mg tid and 20 mg tid were not effective compared with placebo in improving symptoms in NUD patients. This study re-emphasizes the good prognosis of patients with NUD, with 14% of patients improving in the two-week placebo run-in phase and a further 33% improving in the next six weeks while on placebo. Within-treatment analysis of investigator assessments showed improvement for cisapride 20 mg tid suggesting a trend of efficacy at this dose.