Using a wall thickness of greater than 5 mm for the first portion of the duodenum as the criterion for the sonographic diagnosis of duodenal ulcer, we studied the value of sonography in detecting this lesion. Endoscopy (88 patients) and upper gastrointestinal barium studies (12 patients) were used as the gold standards for the diagnosis. The study included 100 patients in whom gastrointestinal disease was suspected (20 with duodenal ulcer and 80 with normal findings). Of the 20 duodenal ulcers verified by endoscopy or upper gastrointestinal series, 13 patients had duodenal wall thickening, six had normal wall thickness, and one was nondiagnostic. Of the 80 subjects with normal findings on endoscopy or upper gastrointestinal series, 73 patients had a normal duodenum, four had wall thickening, and three were nondiagnostic. Considering the four nondiagnostic cases as sonographic errors, there were six false-negative cases and four false-positive cases (a sensitivity of 65%, specificity of 91%, positive predictive value of 76%, and negative predictive value of 92%). Our results show that sonography is not sufficiently sensitive to be used as a screening test, nor is it specific for duodenal ulcer, but a thickened duodenal wall of over 5 mm on sonography warrants additional work-up.