OBJECTIVE: Potential process differences between hospital and community-based endoscopy for Barrett's oesophagus have not been examined. We aimed at comparing adherence to guidelines and neoplasia detection rates in medical centres (MC) and community practices (CP). DESIGN: Retrospective analysis. SETTING: All histologically confirmed Barrett cases seen over a 3-year period in six MC and 19 CP covering a third of all upper gastrointestinal endoscopies (n = 126,000) performed annually in Berlin, Germany. MAIN OUTCOME MEASURE: Rate of relevant neoplasia (high-grade intraepithelial neoplasia or more) in both settings in relation to adherence to standards. RESULTS: Of 1317 Barrett cases, 66% were seen in CP. CP patients had a shorter mean Barrett length (2.6 cm vs. 3.8 cm; P <0.001) with fewer biopsies taken during an examination (2.5 vs. 4.1 for Barrett length