A series of 231 patients with "primary" acute pericardial disease (acute pericarditis or tamponade presenting without an apparent cause) were studied according to the following protocol: general clinical and laboratory studies (stage I), pericardiocentesis (stage II), pericardial biopsy (stage III) and blind antituberculous therapy (stage IV). In 32 patients (14%) a specific etiologic diagnosis was obtained (13 with neoplasia, 9 with tuberculosis, 4 with collagen vascular disease, 2 with toxoplasmosis, 2 with purulent pericarditis and 2 with viral pericarditis). "Diagnostic" pericardiocentesis (32 patients) was performed when clinical activity and effusion persisted for longer than 1 week or when purulent pericarditis was suspected, whereas "therapeutic" pericardiocentesis (44 patients) was performed to treat tamponade; their diagnostic yield was 6% and 29%, respectively. "Diagnostic" biopsy (20 patients) was carried out when illness persisted for longer than 3 weeks, whereas "therapeutic" biopsy was performed whenever pericardiocentesis failed to relieve tamponade; their diagnostic yield was 5% and 54%, respectively. The diagnostic yield difference between "diagnostic" and "therapeutic" procedures was significant (p less than 0.001); in contrast, the global diagnostic yield of pericardiocentesis (19%) and biopsy (22%) was similar. At the end of follow-up (1 to 76 months, mean 31 +/- 20), no patient in whom a diagnosis of idiopathic pericarditis had been made showed signs of pericardial disease. It is concluded that a "diagnostic" procedure is not warranted as a routine method, a choice between "therapeutic" pericardiocentesis and biopsy is circumstantial and must be individualized, and only through a systematic approach can a substantial diagnostic yield be reached in primary acute pericardial disease.