objective: to compare tuberculosis cure rates among patients supervised by household members or health care workers. methods: prospective cohort study of 171 patients treated by the program in vitoria, southeastern brazil, from 2004 to 2007. each patient was followed-up for six months until the end of the treatment. of the patients studied, a household member supervised 59 patients and healthcare workers supervised 112 patients. patients' sociodemographic and clinic data were analyzed. differences between groups were assessed using chi-square test or student's t-test. significance level was set at 5%. results: most patients had smear positive, culture confirmed pulmonary tuberculosis. two patients were hiv-positive. there were more illiterate patients in the healthcare-supervised group, in comparison to those supervised by their families (p=0.01). all patients supervised by a household member were cured compared to 90% of the patients supervised by health care workers (p = 0.024). conclusions: successful tuberculosis treatment was more frequent when supervised by household members.