One hundred and twenty consecutive outpatients with non-ulcer dyspepsia (NUD) and erosive prepyloric changes (EPC) were, after a 2-week placebo run-in period, randomly allocated to double-blind treatment with either 10-mg cisapride tablets or placebo three times daily for 4 weeks. The patients' global evaluation and total symptom score were significantly in favour of cisapride at 2 weeks (p less than 0.05). At 4 weeks the effect of cisapride was no longer significant (p = 0.22). Similarly, the investigators' global evaluation showed marked to moderate symptom improvement in 47% of the cisapride-treated patients as compared with 30% of the placebo-treated patients at 2 weeks. The 95% confidence interval of the difference (18%) was 0% to 35%. At 4 weeks the intergroup difference was only 10% (cisapride, 50% versus placebo 40%). Pain on awakening was the only symptom improved in favour of cisapride at 4 weeks. Thus, when patients with NUD and EPC are treated with cisapride, the therapeutic gain might vanish after the 2nd week of treatment.