Background and aims: While patients with Barrett's oesophagus develop oesophageal adenocarcinoma more frequently than the general population, it has controversially been suggested that gastrooesophageal reflux (GORD) itself is a more important determinant of risk. In order to assess the validity of this suggestion, we examined the risk of oesophageal cancer in patients with Barrett's and with GORD compared with the general population in a community based cohort study. Methods: Cohorts of patients with Barrett's (n = 1677), oesophagitis (n = 6392), and simple reflux (n = 6328), and a reference cohort (n = 13416) were selected from the General Practice Research Database. The last three cohorts were matched to the Barrett's cohort by general practitioner practice, age, and sex. Cox's regression analysis was used to calculate relative risks for oesophageal cancer. Standardised incidence ratio methodology was used to estimate the relative risks for oesophageal adenocarcinoma. Results: A total of 137 oesophageal cancers were identified, of which 94 prevalent cases were excluded. The hazard ratios for oesophageal cancer were 10.6 (5.1-22.0), 2.2 (0.9-5.2), and 1.7 (0.7-4.5) in the Barrett's, oesophagitis, and reflux cohorts compared with the reference cohort, respectively. The corresponding relative risks for oesophageal adenocarcinoma were 29.8 (9.6-106), 4.5 (1.04-19.6), and 3.1 (0.6-14.2). Conclusion: Barrett's oesophagus increases the risk of oesophageal cancer approximately 10 times and oesophageal adenocarcinoma approximately 30 times compared with the general population. There is only a modestly increased risk of oesophageal cancer in patients with reflux who have no record of Barrett's oesophagus. Our findings therefore do not support the suggestion that gastro-oesophageal reflux disease itself predisposes to cancer.