240 patients with active tuberculous pericardial effusion received a 4-drug daily antituberculosis regimen for 6 months and have been studied for 24 months or longer. Those willing were randomly allocated to open pericardial biopsy and complete drainage of pericardial fluid on admission or percutaneous pericardiocentesis as required. All patients were randomly allocated to prednisolone or matching placebo for the first 11 weeks, on a double-blind basis. Complete open drainage on admission abolished the need for pericardiocentesis (p less than 0.01) but did not influence the need for pericardiectomy for subsequent constriction or the risk of death. Among patients who did not have open drainage on admission, 2 (3%) of 76 given prednisolone compared with 10 (14%) of 74 given placebo died of pericarditis (p less than 0.05), 6 (8%) and 9 (12%) respectively required pericardiectomy, 7 (9%) and 17 (23%) repeat pericardiocentesis (p less than 0.05), and 3 (4%) and 7 (9%) open surgical drainage. By 24 months, apart from the 16 who died from pericarditis, all but 3 patients (2%) had a favourable status.