The incidence of coronary perforation using new percutaneous revascularization techniques may be increased compared with PTCA. Still, perforation is uncommonly reported, and the optimal management and expected outcome remain unknown. The objectives of the study were to determine the incidence of coronary perforation using balloon angioplasty (percutaneous transluminal coronary angioplasty, PTCA) and new revascularization techniques and to develop optimal strategies for its management based on classification and outcome. Eleven sites with frequent use of new revascularization devices and prospective coding of consecutive procedures for coronary perforation during 1990 to 1991 contributed to a perforation registry. Patients with perforation were matched by device with an equal-sized cohort without perforation. Data were collected centrally, and all procedural cineangiograms were reviewed at a core angiographic laboratory. A classification scheme based on angiographic appearance of the perforation (I, extraluminal crater without extravasation; II, pericardial or myocardial blushing; III, perforation > or = 1-mm diameter with contrast streaming; and cavity spilling) was evaluated as a predictor of outcome and as a basis for management. Perforation was observed in 62 of 12,900 procedures reported (0.5%; 95% confidence interval, 0.4% to 0.6%), more commonly with devices intended to remove or ablate tissue (atherectomy, laser) than with PTCA (1.3%, 0.9% to 1.6% versus 0.1%, 0.1% to 0.1%; P < .001). The perforation population was notable for its advanced age (67 +/- 10 years) and high incidence of female sex (46%) (both P < .001 compared with patients without perforation). Perforation could be treated expectantly or with PTCA but without cardiac surgery in 85%, 90%, and 44% of class I, II, and III perforations, respectively. Class I perforations (n = 13, 21%) were associated with death in none, myocardial infarction in none, and tamponade in 8%. The incidences of these adverse events were 0%, 14%, and 13% in class II perforations (n = 31, 50%) and 19%, 50%, and 63% in non-cavity spilling class III perforations, respectively (n = 16, 26%). Two of the 15 instances of cardiac tamponade (13%) were delayed, occurring within 24 hours after dismissal from the catheterization laboratory. The incidence of perforation, while low, is increased with new devices. Women and the elderly are at highest risk. The clinical risk after perforation can be classified angiographically, but even low-risk perforations occasionally have poor clinical outcome. Patients should be observed for delayed cardiac tamponade for at least 24 hours.