In a single centre double blind study of 66 patients, the value of cisapride (10 mg twice daily) was compared with placebo in the prevention of duodenal ulcer relapse. Patients who remained ulcer free attended for clinical review every two months and had a mandatory endoscopy at 0, 4, 8, and 12 months or if symptoms suggestive of ulcer recurrence developed. The 12 month crude relapse rates (that underestimate the probability of ulcer recurrence) showed that cisapride was superior to placebo (34% (11/32) relapsed on cisapride v 68% (23/34) on placebo, p = 0.007). This finding was confirmed using lifetable analysis, with a 35% reduction (95% confidence intervals 10-59%, p < 0.05) in the proportion of ulcer relapses in patients who had received cisapride compared with those treated with placebo. These results are similar to those reported in maintenance trials of H2 receptor antagonists analysed by the same method. Drug related adverse clinical events were mainly trivial, but led to three patients on cisapride and one on placebo withdrawing from the trial.