We have studied the accuracy of diagnostic methods in achieving a diagnosis in 75 patients with 81 proven episodes of recurrent ulceration. When the endoscopic findings did not permit an accurate diagnosis, radiology usually provided no additional information. The sensitivity of a combined diagnostic approach was not different from that of endoscopy alone, and so the routine use of upper GI radiographs in addition to endoscopy should, therefore, be abandoned. We also assessed observer variation among endoscopists prospectively in 38 patients investigated for dyspepsia after operation for peptic ulcer. Major observer variation in diagnosing suspected recurrent ulceration occurred in 11% of patients, with a sensitivity and specificity higher than 77 and 94%, respectively. As not all the postoperative gastroduodenojejunal mucosal breaches are peptic, we suggest that acid output and enterogastric reflux assessments are useful diagnostic adjuncts to endoscopy.