A tuberculosis programme in hill and mountain districts of Nepal supported by an international non-governmental organisation (NGO). To evaluate under programme conditions the effectiveness of unsupervised monthly-monitored treatment using an oral short-course regimen. In this prospective cohort study, outcomes for new cases of smear-positive tuberculosis starting treatment over a two-year period in four districts in which a 6-month rifampicin-containing regimen was introduced as first-line treatment (subjects) were compared to outcomes for similarly defined cases in four districts where a 12-month regimen with daily streptomycin injections in the intensive phase continued to be used (controls). Of 359 subjects started on the 6-month regimen, 85.2% completed an initial course of treatment compared to 62.8% of 304 controls started on the 12-month regimen (P < 0.001); 78.8% of subjects and 51.0% of controls were confirmed smear-negative at the end of treatment (P < 0.001). The case-fatality rate during treatment was 5.0% among subjects and 11.2% among controls (P = 0.003). Among those whose status was known at two years, 76.9% of subjects were smear-negative without retreatment, compared to 60.9% of controls (P < 0.001). In an NGO-supported tuberculosis control programme in remote districts of Nepal, patient-responsible short-course therapy supported by rapid tracing of defaulters achieved acceptable outcomes. Where access and health care infrastructure are poor, district-level tuberculosis teams responsible for treatment planning, drug delivery and programme monitoring can be an appropriate service model.