Management strategies in patients with bleeding ulcers are based on ulcer appearance. We assessed the level of agreement among endoscopists regarding the labeling of endoscopic features of bleeding ulcers and also evaluated the effect of a short teaching session on the level of agreement. Two hundred two endoscopists at the 1992 American College of Gastroenterology Postgraduate Course answered multiple-choice questions using interactive keypads. They were shown 6 slides of typical ulcer stigmata of recent hemorrhage, attended a brief teaching session, and then saw 6 different slides of comparable stigmata plus 2 additional slides. Color of stigmata was assessed in 9 slides. Information on years of endoscopic experience and numbers and types of procedures performed was also obtained. The proportion of correct answers before the teaching session increased significantly with years of experience, reaching a plateau in endoscopists who were 6 years or more beyond training. The proportion of correct answers for physicians performing no more than 5 upper gastrointestinal endoscopies per month was significantly lower than for those performing more than 5 per month: 33/56 (59%) compared with 647/891 (73%), a difference of 14% (95% CI, 0.4% to 27%; p = 0.03), but performance of a greater number of endoscopies was not associated with a further increase in the rate of correct answers. The total number of correct answers increased from 845/1181 (72%) before the teaching session to 1268/1554 (82%) afterwards, a difference of 10% (95% CI, 7% to 13%; p < 0.001).(ABSTRACT TRUNCATED AT 250 WORDS)