Thirty-five consecutive patients with massive and submassive pulmonary embolism (PE) were reviewed. In 75% of these cases, PE could be suspected on the basis of electrocardiogram alone. Echocardiography was quite useful for diagnosing PE and assessing right ventricle after-load at the bedside. Spiral computer tomography was effective for obtaining a definitive diagnosis even in a relatively hemodynamically unstable patient. Thrombolysis therapy was given to 30 cases and was apparently effective in 17 cases (17/30, 56.7%). Percutaneous cardiopulmonary support (PCPS) was needed for 7 severe cases. Seven patients, including 5 of the PCPS recipients, underwent surgical embolectomy. Overall mortality was 28.6% (10/35), and surgical mortality was 28.6% (2/7). The significant predictors for mortality were systolic blood pressure <100 mmHg, dopamine >5 microg x kg(-1) min(-1), pH <7.4, PaCO2 >40 torr, base excesss <-5 mmol/L, urine output <0.8ml x kg(-1) x h(-1), intubation, cardiopulmonary resuscitation, duration from attack to emergency room >5h, shock duration >4h, aspartate aminotransferase >100U/L, alanine aminotransferase >100U/L and lactate dehydrogenase >600U/L. Predictors of surgical intervention were dopamine >5 microg x kg(-1) min(-1), shock duration >4h and PCPS. Early initiation of thombolysis therapy is recommended, except in case where it is absolutely contraindicated. Because it is difficult in the early phase of PE to make a decision to perform surgery, aggressive application of PCPS is recommended for the cardiogenic shock associated with massive PE.