13,318 patients admitted to fifty-two coronary care units with suspected acute myocardial infarction were considered for inclusion in a double-blind study comparing recombinant tissue-type plasminogen activator (rt-PA) 100 mg plus heparin with placebo plus heparin. 8307 (62%) were excluded, mainly because their symptoms had begun more than 5 h previously, but all excluded patients were followed up at least until hospital discharge. 2516 patients were randomly allocated to rt-PA and 2495 to placebo. At one month the overall case fatality rates were 7.2% and 9.8%, respectively, a relative reduction of 26% (95% confidence interval 11-39%). 6.3% of patients given rt-PA had a bleeding complication (1.4% major) compared with 0.8% given placebo (0.4% major). However, the incidence of stroke was similar--1.1% in the rt-PA group and 1.0% in the placebo group. Subset analysis showed that patients who had a normal electrocardiogram (ECG) at the time of randomisation (17.5% of the whole trial population) had a low case fatality rate (1.6% in those given rt-PA compared with 3.0% in those given placebo). In those with an abnormal ECG at entry, rt-PA was associated with a 24.5% relative reduction in 1 month fatality (95% confidence interval 9-37%).