Endoscopic follow-up study of gastric ulcer to detect cancer is held to be mandatory. To evaluate the benefits of this routine strategy, 913 index endoscopies for gastric ulcer in 773 patients during the 3-year period 1985-87 were analyzed. Correctness of diagnosis was verified through surgery, autopsy, or clinical follow-up. Endoscopic follow-up was done in 83% of the cases, totaling 1269 endoscopies, showing gastric cancer in 10 patients. Clinical outcome, however, was poor for five of these (early death). Five additional cancer cases were missed by the endoscopic follow-up. In the same period 63 gastric cancers were found at the first endoscopy; 9 of these were diagnosed through biopsies only. Predictive values of the macroscopic judgements of benign lesion or probable/definite malignancy were 0.98 and 0.40, respectively. Evaluation of case records did not indicate characteristics that would have helped in the correct differentiation between benign and malignant lesions. Thus, each case of curable gastric cancer is found at the expense of approximately 250 follow-up endoscopies. We are in need of sensitive and specific markers for possible malignancy in the patient with apparently benign gastric ulcer.