Major upper gastrointestinal bleeding (UGIB) is the most frequent serious side effect of aspirin and other nonsteroidal anti-inflammatory drugs (NSAIDs). An existing case-control study mounted to quantify the risk of UGIB among users of different analgesics and NSAIDs (and other factors) will be extended. Cases are patients admitted to hospitals in eastern Massachusetts with first episodes of UGIB, without antecedent peptidic ulcer disease, cirrhosis, or other predisposing conditions. Neighbor controls of the same age and sex as the cases are identified from town lists at a ratio of two per case. Study subjects are interviewed at home by telephone to obtain information on analgesic and other drug use in the preceding six months, personal data and habits (e.g., marital status, cigarette smoking, alcohol consumption), and relevant medical history. It is expected that at least 490 cases of gastric bleeding, 245 cases of duodenal bleeding, and 1470 controls will be enrolled during the study extension, increasing the available numbers to about 750, 345, and 2190, respectively. The data will be used to quantify associations between individual analgesics and NSAIDs and UGIB, and to compare the relative effects of the different drugs. The effects of different doses will be evaluated. (The relation of UGIB to the use of other factors, in particular cigarette smoking, and alcohol and coffee consumption, will also be investigated.) Specific issues with drugs to be focused on in the study extension include the effect of aspirin taken regularly at very low doses (an interim analysis from the present study suggests that the use of 325 mg or less at least every other day quadruples the risk of gastric bleeding, which raises the question about even lower doses), the effect of ibuprofen at different doses, and the effect of diclofenac, recently introduced to the U.S. market. Absolute risks attributable to these and other NSAIDs will be calculated after deriving an estimate of the incidence of UGIB from data provided by the Massachusetts Health Data Consortium.