OBJECTIVE--To examine the association between dyspeptic symptoms and endoscopic and histological diagnoses. DESIGN--Cross sectional study of people with dyspepsia and controls matched for age and sex identified by questionnaire survey of all inhabitants aged 20 to 69. Endoscopy and histological examination was performed with the examiner blind to whether or not the patient had dyspepsia. SETTING--Population based survey in Sørreisa, Norway. SUBJECTS--All people with dyspepsia and age and sex matched people without dyspepsia were offered endoscopy. A total of 309 people with dyspepsia and 310 without dyspepsia underwent endoscopy, giving 273 matched pairs. MAIN OUTCOME MEASURES--Prevalences of endoscopic and histological diagnoses made according to internationally accepted standards. RESULTS--In all, 1802 of 2027 (88.9%) people returned the questionnaire. Of the 163 subjects who refused endoscopy, 114 were controls. Of five endoscopic and four histological diagnoses only peptic ulcer disease, endoscopic duodenitis, and active chronic gastritis were diagnosed significantly more often in people with dyspepsia. In all, 30% to 50% of the diagnoses of mucosal inflammation and peptic ulcer disease were made among subjects without dyspepsia, and only 10% of both those with and those without dyspepsia had normal endoscopic findings. CONCLUSIONS--The diagnostic findings, with possible exceptions of peptic ulcer disease and endoscopic duodenitis, showed no association of clinical value with dyspeptic symptoms. The small number of "normal" endoscopic findings in both those with and those without dyspepsia challenge well accepted endoscopic and histological diagnostic criteria with relation to the upper gastrointestinal tract.