To evaluate the association between nonsteroidal antiinflammatory drug (NSAID) use and upper gastrointestinal bleeding (UGIB) and lower gastrointestinal bleeding (LGIB), we performed a prospective case-control study at a large inner-city hospital over a 28-month period evaluating 461 consecutive patients hospitalized for UGIB and 105 with LGIB. During the same period, 1895 in-patients evaluated by our gastroenterology consultative service served as controls. At the time of initial evaluation, all patients were asked about the use of any prescription or over-the-counter NSAID product within one week of admission. Endoscopic examination was performed in most patients with bleeding. NSAID use was almost equivalent in patients with UGIB and LGIB (60%) and significantly greater than controls [34%; P < 0.001; odds ratio (OR) 3.0; 95% CI, 2.4-3.6]. The age, race, and gender adjusted risk for LGIB associated with NSAID use was significant [adjusted OR (AOR) 2.6; 95% CI 1.7-3.9], although less than UGIB (AOR 3.2; P = 0.34). The risk associated with diverticular bleeding (N = 53, AOR 3.4; 95% CI 1.9-6.2) was higher than duodenal ulcer bleeding although not significantly (N = 97, AOR 3.0). We conclude that NSAID use is strongly associated with LGIB and from lesions not considered associated with mucosal ulceration such as diverticulosis.