The aim of our prospective randomized study involving 100 patients was to investigate whether Doppler ultrasound can be used to select patients at risk for ulcer rebleeding. Ulcers in the Forrest group classified as having a visible vessel or a clot were treated prophylactically by injection with epinephrine solution. In the Doppler group, in contrast, only ulcers with a positive Doppler signal were treated endoscopically. In the Doppler group, rebleeds occurred significantly less frequently (2%, P < 0.03) than in the Forrest group (14%). Emergency surgery was only necessary in the Forrest group (0% vs 5%; P = 0.02). Bleeding-related mortality was 0% and 4% (P = 0.15) and the overall mortality 0% and 10% (P = 0.02), in the Doppler and Forrest groups, respectively. These results appear to show that Doppler-based injection treatment is superior to endoscopic treatment based exclusively on the Forrest classification. In our study, Doppler-based local endoscopic treatment reduced the danger of a rebleed and thus the number of emergency operations and the overall mortality.