The aim of the systematic review was to determine the effect of cisapride compared with placebo or other non-surgical therapies for the treatment of symptoms of gastro-oesophageal reflux in children. We searched MEDLINE, EMBASE, the Cochrane Controlled Trials Register, Science Citation Index and reference lists for randomized controlled trials which compared cisapride with placebo or other non-surgical therapy in children. We included only trials which reported reflux-related symptoms as an outcome, provided that cisapride was administered orally for at least I week. Seven trials (286 children in total) compared cisapride with placebo. Two trials reported good concealment of treatment allocation. The pooled odds ratio for the 'same or worse' symptoms was 0.34 (95% CI 0.10, 1.19). There was substantial heterogeneity between studies (P < 0.00001) and the funnel plot was asymmetrical. Adverse effects (mainly diarrhoea) were not significantly increased with cisapride (pooled odds ratio (OR) 1.80: 0.87, 3.70). The reflux index was significantly reduced in children treated with cisapride (weighted mean difference -6.49: -10.13, -2.85). One study (50 children) compared cisapride with gaviscon plus carobel: the OR for the 'same or worse' symptoms was 3.26 (0.93, 11.38). There was no clear evidence that cisapride reduced symptoms of gastro-oesophageal reflux. As smaller, poorer quality studies were biased in favour of a positive treatment effect, the pooled OR overestimated the potential benefits of cisapride. There was some evidence to suggest that gaviscon plus carobel may be a more effective option than cisapride.