A survey was undertaken in April 1993 to compare the respective benefits of 2 regimens containing either streptomycin (SHRZ) or ethambutol (EHRZ) in the first two months of treatment of smear-positive pulmonary tuberculosis in Madagascar. This operational research was justified by the risks related to the use of parenteral streptomycin in a country where single use material is rare and its purpose was to provide arguments for an eventual recommendation to replace this drug by oral ethambutol which is also less expensive. 907 patients were included. The compliance was not significantly different between the 2 groups, although it was traditionally assumed to be better with streptomycin. The frequency of side effects was significantly lower with EHRZ. Overall treatment failure rates were not significantly different, but all of 6 patients who were negative at 5 months and were again positive at 8 months had received EHRZ. This point obliged to be careful before concluding, because 24% of patients were lost for follow-up. A 2 years surveillance will be necessary to compare the frequency of recurrences.