The relationship between upper gastrointestinal hemorrhage and drug use was studied in 251 Chinese patients (179 men, 72 women) admitted to the Prince of Wales Hospital, Hong Kong, and control subjects matched for age and sex. There was a highly significant difference between the cases and control subjects in the use of NSAIDs (odds ratio 14.0, p < 0.00001), ulcer healing drugs (odds ratio 12.5, p < 0.00001), and Chinese proprietary medicines (odds ratio 16.0, p < 0.00001). There was also a significant difference in the use of analgesics (odds ratio 14.0, p = 0.001), paracetamol (odds ratio 2.5, p = 0.01), antacids (odds ratio 2.7, p < 0.001) and unknown drugs (odds ratio 4.7, p < 0.001). Cases also differed from control subjects regarding the use of tobacco (odds ratio 2.3, p < 0.001) and alcohol (odds ratio 1.7, p = 0.02), and the presence of peptic ulcer symptoms (odds ratio 29.8, p < 0.00001). Significantly more control subjects than cases were receiving aspirin, cardiovascular drugs, bronchodilators, oral hypoglycemic drugs/lipid-lowering drugs, and anticonvulsants/hypnotics, due to the inevitable differences in disease pattern between the 2 groups. NSAID use was a major factor associated with upper gastrointestinal hemorrhage from primarily peptic ulcers. Differences in the use of other drugs may reflect variations in disease patterns between cases and controls, the common practice of self-medication in Hong Kong, and the concomitant use of NSAIDs and ulcer healing drugs/antacids.