BACKGROUND: The incidence of distal oesophageal adenocarcinoma is rising, with chronic reflux and Barrett's oesophagus being considered risk factors. Reliable detection of Barrett's oesophagus during upper endoscopy is therefore mandatory but requires both endoscopy and histology for confirmation. Appropriate management of patients with endoscopic suspicion but negative on histology, or vice versa, or of patients with no endoscopic suspicion but with a biopsy diagnosis of intestinal metaplasia at the gastro-oesophageal junction, has not yet been studied prospectively. PATIENTS AND METHODS: In a prospective multicentre study, 929 patients (51% male, mean age 50 years) referred for upper gastrointestinal endoscopy were included; 59% had reflux symptoms. The endoscopic aspect of the Z line and any suspicion of Barrett's oesophagus were noted, and biopsies were taken in all patients from the Z line (n = 4), gastric cardia (n = 2), and body and antrum (n = 2 each). Biopsies positive for specialised intestinal metaplasia (SIM) were reviewed by a reference pathologist for a final Barrett's oesophagus diagnosis. All patients with endoscopic and/or histological suspicion of Barrett's oesophagus were invited for a follow up endoscopy; the remaining cases (no endoscopic or histological suspicion of Barrett's oesophagus) were followed clinically. RESULTS: Of 235 patients positive for Barrett's oesophagus on endoscopy and/or histology, 63% agreed to undergo repeat endoscopy (mean follow up period 30.5 months). 46% of patients with an endoscopic Barrett's oesophagus diagnosis but no histological confirmation (group A) showed the same distribution, a further 42% did not have Barrett's oesophagus, and 11% had confirmed Barrett's oesophagus on both endoscopy and biopsy on follow up. In the group with a histological Barrett's oesophagus diagnosis but negative on initial endoscopy (group B), follow up showed the same in 26% whereas 46% had no Barrett's oesophagus, and confirmed Barrett's oesophagus (endoscopy plus histology) was diagnosed in 17%. Of the study population, 16 patients had Barrett's oesophagus on initial endoscopy confirmed by histology which remained constant in 70% at follow up (group C). Of the remaining patients without an initial Barrett's oesophagus diagnosis on either endoscopy or histology (group D) and only clinical follow up (mean follow up period 38 months), one confirmed Barrett's oesophagus case was found among 100 patients re-endoscoped outside of the study protocol. However, no single case of dysplasia or cancer of the distal oesophagus was detected in any patient during the study period. CONCLUSIONS: Even in a specialised gastroenterology setting, reproducibility of presumptive endoscopic or histological diagnoses of Barrett's oesophagus at follow up were poor. Only 10-20% of cases with either endoscopic or histological suspicion of Barrett's oesophagus had established Barrett's oesophagus after 2.5 years of follow up. The risk of dysplasia in this population was very low and hence meticulous follow up may not be required.