The diagnostic value of biphasic radiographic examination of the stomach and duodenum was compared with that of fiberoptic endoscopy in a prospective, blinded study of 385 patients with dyspepsia. This investigation was directed at gastric malignancies and peptic ulcers. Methodologically there is no absolute standard for a study of this kind because histologic examination is useful for detection of cancer but inadequate for ulcers. As an alternative, kappa indexes and the sensitivity and specificity, as derived by Hui and Walter, were calculated and compared. For the detection of gastric carcinoma, radiographic and endoscopic findings had almost perfect agreement beyond chance. For gastric ulcers, radiography and endoscopy had substantial agreement, which became perfect if small ulcers (less than 5 mm) were excluded. For duodenal ulcers, radiography had a lower sensitivity than endoscopy; this disagreement disappeared if small ulcers were excluded. Both methods have equal merit; choice of the initial diagnostic procedure will therefore depend on cost, discomfort to the patient, and risk of complications.