A prospective controlled study was carried out to determine the validity of the Forrest classification in terms of improved laser therapy. Out of 153 consecutive patients with bleeding peptic ulcers, 137 patients--74 with arterial ulcer bleeding and 63 with non-arterial ulcer bleeding--were included in the trial. In arterial ulcer bleeding a significantly lower rate of permanent hemostasis was achieved, and the frequency of urgent surgery and mortality was higher than in non-arterial ulcer bleeding. Patients with spurting arterial bleeding (Forrest Ia) and those with large non-bleeding visible vessels (Forrest IIa) include almost all patients at risk of further bleeding and death. Combined laser therapy clearly minimizes the risk of rebleeding and death in small non-bleeding visible vessels (Forrest IIa). In contrast, decreased hemoglobin and a requirement for blood transfusion are of limited prospective value for the individual emergency patient. Overall, our results demonstrate that Forrest criteria are essential for proper planning of endoscopic therapy and urgent surgery in bleeding peptic ulcers. Emergency endoscopy must therefore be performed as early as possible.