Little is known about the aetiology of Barrett's oesophagus, a condition in which columnar epithelium replaces normal squamous epithelium above the juncture of the oesophagus with the stomach. An exploratory retrospective study of males from the Tucson Veteran's Affairs Medical Center (Tucson, AZ, USA), looking at 22 Barrett's cases and 22 controls, was undertaken to: (1) determine whether risk factors previously reported for squamous oesophageal cancer were also potential risk factors for Barrett's oesophagus; and (2) explore which of several retrospectively determined dietary factors would permit maximum discrimination of Barrett's cases from controls. Because of the recognized difficulty with retrospective dietary assessment, an intensive interview-based life history approach was developed to assess behavioural factors throughout adulthood. We examined the dose-response relationship involving questions (1) and (2) separately for subjects with lesions less than or greater than 7 cm in length. A literature-based oesophageal cancer risk factor score (including use of tobacco and intakes of alcohol, fat, vitamin A, linoleic acid and protein) discriminated cases from controls. Alcohol consumption alone discriminated patients with lesions < 7 cm in length from controls, and a more complex risk factor score (including tobacco use, alcohol, fat, fruits and vegetables) discriminated patients with lesions > 7 cm in length from controls.