Background: Most dyspeptic patients in primary care are managed without confirmatory investigations. In this study the reliability of the unaided clinical diagnosis and the diagnostic value of dyspepsia subgrouping are evaluated in unselected dyspeptic patients in primary care. Methods: Six hundred and twelve unselected dyspeptic patients were referred for interview and endoscopy. General practitioners stated a provisional diagnosis and a proposed management strategy. Before endoscopy, patients were classified on the basis of predominant symptoms as reflux-, ulcer-, or dysmotility-like or as unclassifiable Results: The sensitivity and the positive predictive value of the diagnosis of ulcer were 0.58 and 0.29, respectively, and those for esophagitis 0.30 and 0.43. The predictive value of a clinical diagnosis of functional dyspepsia was high, but, considering the high prevalence of the condition, the chance-corrected validity was at the same level as for the other diagnoses (0.18-0.22). Classification of patients by predominant symptoms increased the a priori probability of ulcer and ersophagitis in the respective subgroups. However, more than one-third of the patients with ulcer or esophagitis were classified in inappropriate subgroups. Conclusions: It is difficult to select an appropriate management strategy for dyspeptic patients on the basis of symptoms and history alone. Dyspepsia subgroups are of limited help in the decision process because of the low predictive value of the endoscopic diagnosis.