Five 6-month antituberculosis regimens, allocated at random to patients with acid-fast bacilli in their sputum on microscopy, were studied. Four, given three times a week from the start, contained isoniazid and rifampicin together with (1) streptomycin, pyrazinamide, and ethambutol, (2) streptomycin and pyrazinamide, (3) streptomycin and ethambutol, or (4) pyrazinamide and ethambutol. The fifth was daily isoniazid, rifampicin, pyrazinamide, and ethambutol. All 833 patients with drug-sensitive strains of bacilli before treatment had a favourable bacteriological response during chemotherapy, and the bacteriological relapse rates during 12 months after stopping chemotherapy were 2% or less for all regimens except thrice-weekly isoniazid, rifampicin, streptomycin, and ethambutol (the only regimen without pyrazinamide), which had a relapse rate of 8%. The results were equally good for the 138 patients with bacilli resistant to isoniazid, streptomycin, or both drugs initially. The incidence of potentially serious toxicity was low. The daily regimen is relevant to programmes in which patients self-administer their drugs, and the 3 pyrazinamide-containing intermittent regimens are relevant to fully supervised outpatient programmes.