Two chemotherapy regimens of 28 weeks (6M) and 36 weeks (8M) duration were compared in patients with previously untreated pulmonary tuberculosis in 30 health districts of Algeria. In the 6M regimen, patients received isoniazid plus rifampicin daily throughout with streptomycin and pyrazinamide in addition for the first 8 weeks. In the 8M regimen patients received isoniazid daily throughout plus rifampicin for the first 16 weeks and streptomycin and pyrazinamide as well for the first 8 weeks. The districts were allocated to apply either the 6M or the 8M regimen to all eligible patients. However, 22% of eligible patients were not admitted, the most common reason being oversight on the part of the personnel in the local health centres. A total of 2218 (977 6M, 1241 8M) patients was admitted and the results were assessed 2 years after the due date of completion of chemotherapy, irrespective of whether or not management had deviated from the study protocol. An appreciable proportion of patients (25% of 6M, 29% of 8M) could not be assessed because they failed to attend for follow-up or the collection of specimens was overlooked or because of difficulties in the transport of specimens to the reference laboratory in Algiers, and a further 5% and 4% of patients died. Of those who were assessable 7 (1%) of 731 6M and 15 (2%) of 875 8M were classified as having an unfavourable response during or after chemotherapy because they had one or more positive cultures. In addition 16 (2%) 6M and 18 (2%) 8M patients had received additional chemotherapy for failure, relapse or the development of non-pulmonary lesions but had achieved a favourable status. Of 1018 (488 6M, 530 8M) patients assessed who had fully sensitive strains pretreatment and completed treatment as planned, 97% in both series had a favourable status 2 years after their chemotherapy was due to be completed. In patients with initial strains resistant to isoniazid the results were less good in the 8M than the 6M series: none of the 19 6M compared with 5 of 26 in the 8M series having an unfavourable outcome suggesting that the duration of rifampicin may have been important in patients with initial isoniazid resistance. A slightly higher proportion of patients on the 8M regimen defaulted during chemotherapy (9.3% compared with 6.7% in the 6M regimen) but there was no difference between the regimens in other measures of compliance.(ABSTRACT TRUNCATED AT 400 WORDS)