In 1998, the Damien Foundation Bangladesh invited semi-qualified, private "gram dakter" (Bangla for "village doctors") to participate in tuberculosis (TB) programmes in a population of 26 million people in rural Bangladesh. The organization trained 12 525 village doctors to not only refer suspected TB cases for free diagnosis but also to provide directly observed treatment (DOT) free of charge. Source of referral and place of DOT was recorded as part of the standardized TB recording and reporting system, which enabled us to quantify the contribution of village doctors to case detection rates and also allowed disaggregated cohort analysis of treatment outcome. During 2002 and 2003, 11% of all TB cases with positive sputum smears in the study area had been referred by village doctors; the rate of positive tests in patients referred by village doctors was 14.4%. 18 792 patients received DOT from village doctors, accounting for between 20% and 45% of patients on treatment during the 1998-2003 period. The treatment success rate was about 90% throughout the period. Urine samples taken during random checks of treatment compliance were positive for isoniazid in 98% of patients treated by village doctors. Within the framework of Public-Private Mix DOTS, services provided by semi-qualified private health care providers are a feasible and effective way to improve access to affordable high quality TB treatment in poor rural populations. The large informal health workforce that exists in resource poor countries can be used to achieve public health goals. Involvement of village doctors in TB control has now become national policy in Bangladesh.