To assess the relative rate or upper gastrointestinal (UGI) tract bleeding associated with nonsteroidal anti-inflammatory drugs (NSAIDs), we performed a retrospective cohort study using 1980 billing data from all Medicaid patients in the states of Michigan and Minnesota. The rate of UGI tract bleeding in the 30 days following each drug exposure was examined in the 88,044 patients dispensed only one of seven NSAIDs. The rate of UGI tract bleeding differed significantly among users of these drugs. Stratification and logistic regression were used to adjust for multiple potential confounding factors, without substantive changes in the results. An alcohol-drug interaction was found. Sulindac users had the highest rate of UGI tract bleeding, and it was the only drug statistically different from ibuprofen. When the average daily dose of sulindac received was divided by the maximum recommended daily dose, it was notably higher than those for other drugs. Repeated analyses using data from 1982 confirmed these results. We conclude that there are significant and consistent differences in the incidence of UGI tract bleeding associated with the use of NSAIDs in this population.