The risk and potential risk factors of pericardiocentesis were assessed by a review of a series of 52 pericardiocenteses comprising all those performed in the cardiac catheterization laboratory of one institution from 1971 to 1978. On the basis of the operative results, the patients were separated into two groups for comparison; Group I comprised all patients with a successful uncomplicated (35) pericardiocentesis and Group II all those with a nonproductive (16), nontherapeutic (1) or complicated (8) pericardiocentesis. Complications consisted of one death, one cardiac arrest, one aspiration of a subdiaphragmatic abscess and five ventricular punctures without adverse sequelae. Among the patients who had a nonproductive pericardiocentesis, the condition of 11 had probably been misdiagnosed but at least 4 had a false negative pericardiocentesis. Comparison of the two groups showed no significant difference in the incidence of cardiac tamponade or in the clinical presentation based on historical, physical, electrocardiographic, roentgenographic or echocardiographic findings of pericardial disease. Pericardiocentesis was usually successful when performed for suspected malignant pericardial effusion but often unsuccessful when performed for suspected hemopericardium. Anatomically, all patients in Group II had either minimal or loculated posterior pericardial effusion. It is concluded that pericardiocentesis can be performed at a low risk that can be further minimized by consideration of the disease process and the anatomic location of the pericardial fluid.