Fifty patients (38 men) with unstable angina pectoris defined by: pain lasting > 15 minutes+percritical electrocardiographic changes+significant coronary narrowing on coronary angiography (Coro 1) performed within 24 hours, were treated in a double blind protocol with rt-PA (n = 25) 100 mg/90 minutes (10 mg bolus + 90 mg/90 minutes or placebo (n = 25). All received effective intravenous heparin and intravenous nitrates. Calcium antagonists and betablockers were prescribed in half the cases. Aspirin (100 mg orally per day) was prescribed after control coronary angiography (Coro 2) performed 24 +/- 6 hours after starting treatment. Qualitative and quantitative analysis (CAESAR system) was centralised. There were no differences in the angiographic findings between the two groups. Intracoronary thrombosis was observed in 43% (rt-PA) and 44% (placebo) in Coro 1 and in 17% and 28% in Coro 2. The incidence of myocardial revascularisation procedures was similar in the two groups: angioplasty: 12 (rt-PA) and 13 (placebo); coronary bypass surgery: 5 (rt-PA) and 6 (placebo). Seven patients developed myocardial infarction (5 rt-PA, 2 placebo), one of whom died of cardiogenic shock (placebo). Eighteen patients had haemorrhagic complications (14 rt-PA, 4 placebo; p < 0.002) mainly at the puncture sites (12/14, 3/4). Spontaneous haemorrhage occurred in 7/25 (28%) of patients on rt-PA (haematuria 3, gastrointestinal haemorrhage 2, haematuria + gastrointestinal haemorrhage 1, epistaxis 1) and in 1/25 patients on placebo (gastrointestinal haemorrhage) This study shows that intravenous thrombolysis with rt-PA in severe unstable angina pectoris doe not modify the clinical outcome or the angiographic lesions but exposes patients to a high risk of haemorrhagic complications.