A prospective study was undertaken to determine the effect of anticoagulation therapy on the prevalence of occult gastrointestinal bleeding and to ascertain the causes of bleeding. During a six-month period, 256 patients who had received anticoagulants were screened for occult bleeding with guaiac-impregnated cards. Twenty-one (12 percent) of the 175 patients who had received anticoagulants and who voluntarily completed the cards had occult gastrointestinal bleeding compared with only two (3 percent) of 74 control patients who had not received anticoagulants. The mean prothrombin time and partial thromboplastin time did not differ significantly in the patients who had received anticoagulants with and without subsequent bleeding. Fifteen of 16 patients who had received anticoagulants and who underwent diagnostic evaluation had previously undiagnosed lesions in the intestinal tract. These results indicate that occult gastrointestinal bleeding in a patient who has received an anticoagulant should not be attributed to the anticoagulant; instead, such bleeding often indicates the presence of significant intestinal disease.