OBJECTIVES: To extract outcome data on the probability of symptom resolution for acute drug treatment and symptomatic relapse for long-term drug treatment in patients with gastro-esophageal reflux disease (GERD), by drug, dose and by patient category (non-endoscoped GERD, endoscopy-verified reflux esophagitis and GERD without esophagitis). The results were intended for use in cost-effectiveness evaluations, where data on esophagitis healing or remission have traditionally been used as a definition of success in the treatment of GERD. METHODS: A systematic literature search for randomised controlled trials of medical interventions for GERD was carried out in September 2000 by a contract research organisation, the Evidence Research Unit, UK. The citations uncovered underwent a series of filtering steps for applicability and methodological quality. Where possible, extracted data were pooled. RESULTS: Very few of the 82 identified studies were found to contain relevant data as a basis for the symptomatic probabilities required. Reliable pooled data were only found for complete resolution of heartburn after 4 weeks' acute treatment with omeprazole 20mg od in patients with esophagitis (60.9%, n=1,233) and in patients with GERD without esophagitis (63.5%, n=525, p>0.05 for the difference between groups). Only one study was identified concerning the 6-month probability of symptomatic relapse during no drug treatment (i.e., not placebo) in GERD patients with esophagitis (90%, n=145) and without esophagitis (75%, n=123, p<0.001 for the difference between groups). CONCLUSIONS: Reliable symptomatic probabilities intended for cost-effectiveness evaluations of medical interventions for GERD could not be sufficiently obtained in this systematic review of randomised controlled trials.