From July 1988 to April 1993, 18 patients with active, marginal ulcer bleeding were investigated for their clinical presentation, endoscopic findings, and successful rate of therapeutic endoscopy. Through endoscopic survey, data on the ulcer number, ulcer location, and prevalence rate of each major stigmata of recent hemorrhage were obtained. Therapeutic methods used include heater probe thermocoagulation and local injection with diluted epinephrine (0.01%) or 95% alcohol through the endoscope. All patients were male with a mean age of 65.3 years. Tarry stool passage (15/18; 83.3%) was the most common clinical presentation. Fifteen patients (15/18; 83.3%) had single marginal ulcer. Ulcerations were located in the jejunal site (12/18; 66.6%) and gastro-jejunal anastomosis (5/18; 27.7%). There were two patients with spurting artery, six with nonbleeding visible vessel, two with blood clot adhesion, four with oozing, three with visible vessel-associated oozing and three, with blood clot adhesion accompanying oozing. The rate of therapeutic endoscopy to successfully stop bleeding or prevent rebleeding was 94.4% (17/18). Sixteen cases (16/17) received one session of endoscopic therapy, and one patient (1/17) accepted two sessions. One patient with multiple underlying diseases died of persistent bleeding after three sessions of therapy. No complication was found to be related to the endoscopic procedure. Marginal ulcer bleeding exhibits the same endoscopic findings as peptic ulcer hemorrhage. Most marginal ulcers are single and located at the saddle area of the jejunal site. Endoscopy can be useful in diagnosing marginal ulcer, and it is safe and effective in managing active marginal ulcer bleeding.