This article reviews the pathophysiology and management of bleeding peptic ulcer. Ulcers bleed when and because they erode into a blood vessel, and bleed massively when they erode into a medium- or large-sized artery. Focal pathology at the bleeding point (such as arteritis, aneurysmal dilatation or recanalized thrombus) contributes to the timing and clinical pattern of ulcer bleeding. Big bleeds are probably associated with erosion into big arteries. The identification of a visible vessel in the floor of an ulcer that has recently bled is predictive of further bleeding, while the absence of a visible vessel or stigmata of recent haemorrhage is strongly predictive that further bleeding will not occur. Unfortunately, no conventional method of managing gastrointestinal bleeding from ulcers has been convincingly shown to be better than placebo in controlled clinical trials. The value of transfusion and surgery has never been tested in controlled trials, while many small studies of drug therapy, especially of H2-receptor blocking agents, and a few small studies of early surgery afford generally negative or equivocal results. There is some evidence that new physical methods such as lasers or bipolar probes applied at endoscopy are superior to placebo though negative trials have also been reported. Studies randomizing larger numbers of patients with bleeding ulcers are required if therapeutic benefit of any aspect of management is to be demonstrated or refuted.