The risk of bleeding associated with intracoronary infusion of streptokinase in acute myocardial infarction was determined in a randomised controlled trial containing 302 patients under the age of 70. Intracoronary streptokinase infusion was given to 152 patients and 150 patients were treated conventionally. Bleeding was seen in 24 (16%) patients in the streptokinase group and in two of the conventionally treated patients. Bleeding was most common (28%) in patients over the age of 60 years. The groin was the site of bleeding in all patients except one. In the first 48 hours after admission the haematocrit in streptokinase treated patients with manifest bleeding fell by 0.07 (0.04) (mean (SD)). The fall in haematocrit in the streptokinase treated patients without manifest bleeding was 0.05 (0.04) and in the conventionally treated patients it fell by 0.03 (0.04). Sixty six units of packed cells were transfused in the streptokinase group (50 units to those who bled); the control group required only 17 units. There were no deaths due to bleeding. The occurrence of bleeding and the fall in haematocrit in the streptokinase group correlated with the occurrence of systemic fibrinolysis but not with the dose of streptokinase given. Thus, in about 15% of patients treatment with intracoronary streptokinase resulted in significant non-fatal bleeding from the femoral puncture site that required substantial transfusion support. Furthermore, there was a significant drop in haematocrit in patients without manifest bleeding. These results emphasise the need for more specific fibrinolytic agents.