Percutaneous cardiopulmonary bypass (PCB) was instituted in 30 initially stable patients who developed either cardiac arrest refractory to resuscitation (n = 7) or cardiogenic shock (mean arterial blood pressure < 50 mm Hg unresponsive to fluid resuscitation or vasopressors) (n = 23) after a catheterization laboratory complication. Events leading to collapse included abrupt closure during percutaneous transluminal coronary angioplasty (PTCA) (n = 22), complications from diagnostic cardiac catheterization (n = 6), left ventricular perforation during mitral valvuloplasty (n = 1), and right ventricular perforation during pericardiocentesis (n = 1). PCB was initiated within 20 minutes of cardiovascular collapse in 83% of patients (arrest: 21 +/- 13 minutes [range 10 to 50]; and shock: 17 +/- 6 minutes [range 10 to 30]). Mean arterial blood pressure increased on PCB from 0 to 56 mm Hg in patients with cardiac arrest and from 37 to 63 mm Hg in those with cardiogenic shock at mean PCB flow rates of 2.5 to 5.0 liters/min. Subsequent therapy on PCB included emergent cardiac surgery (n = 14), PTCA (n = 13) and medical therapy (n = 3). Six patients (20%) survived to hospital discharge (3 with cardiac surgery, 2 with PTCA, and 1 with medical therapy). All 7 patients with refractory cardiac arrest died despite further interventions on PCB, whereas 6 of 23 (26%) with cardiogenic shock survived to hospital discharge. Thus, in response to cardiovascular collapse in the catheterization laboratory, PCB does not salvage patients who do not regain a stable cardiac rhythm. PCB can stabilize patients who develop cardiogenic shock for further interventions which are lifesaving in only a minority of patients.