We performed a pathological study of experimental canine myocardial infarction (MI) induced by coronary thrombosis which was made by endothelial denudation and induction of luminal stenosis in 20 dogs (Group I). Another model of experimental MI by coronary ligation was also evaluated in Groups II and III. Thirteen dogs of Group I and 7 of Group II underwent persistent coronary occlusion for 6-8 h (Group IA and Group IIA), and 7 of Group I and 7 of Group II underwent coronary reperfusion with intravenous urokinase (UK) (20,000 IU/kg) in Group IB for 5 h following temporary coronary occlusion for 3 h (Groups IB and IIB). The remaining 5 dogs underwent coronary reperfusion for 5 h following intravenous 20,000 IU/kg UK after 3 h ligation. Microscopically, myocardial hemorrhage was present in 6 (86%) Group IB, 4 (31%) Group IA and in no Group IIA dogs (p less than .025 and p less than .005 vs. IB). Four Group IIB and 3 Group III dogs also showed myocardial hemorrhage. Moderate hemorrhage was present only in Group I and slight hemorrhage was frequently observed in reperfused hearts. Contraction band necrosis (CBN) was present in 8 (62%) Group IA, all Group IB, all Group IIB, and 4 (80%) Group III dogs. However, there was no hemorrhage with CBN in Group IIA (p less than .005 vs. IIB). Marked CBN was present only in Group I.(ABSTRACT TRUNCATED AT 250 WORDS)