The proven benefit of aspirin in the secondary prevention of cardiovascular disease and its possible value in primary prevention must be weighted against its potential hazards. This paper is an overview of the gastrointestinal toxicity of aspirin, its most serious complication after intracerebral haemorrhage. Information on toxicity has been drawn only from randomised trials, thus avoiding the potential biases of observational studies. All randomised placebo controlled trials listed in the Anti-platelet Trialists Collaboration where a direct aspirin-placebo comparison was possible were included. Twenty-one trials were included, all but one of secondary prevention. There were over 75,000 person years of aspirin exposure. The pooled odds ratios for categories of gastrointestinal bleeding (e.g. haematemesis, melaena) were between 1.5-2.0; fatal bleeds were very rare. The risk of peptic ulcers was 1.3 and of upper gastrointestinal symptoms 1.7. These risks were lower than those found in observational studies. Attributable disease rates are also presented. For haematemesis for example they varied from 0.2-1.0 per 1000 person years. Toxicity was dose related. Aspirin does have significant gastrointestinal toxicity, although this is rarely fatal. More recent work has demonstrated the efficacy of low doses of aspirin (75 mg daily) but there is limited information yet available on its toxicity.