Setting: Philanthropic health center specialized in lung diseases, including tuberculosis. Background: Directly-observed treatment (DOT) is the chief measure to insure improvement in cure levels and to reduce acquired resistance. This unit adopted DOT in the past, but financial difficulties prevent its maintenance. Objectives: Compare abandon levels in DOT and auto-administered treatment (AT) and identify predictive factors associated to abandonment. Type of study: Retrospective; patients observed in different years and selected at random. Material and methods: 1,226 patients were selected (613 patients in DOT and 613 in AT). Abandonment was compared in DOT and TA; in each group, abandonment was compared to sex, age, alcoholism, previous treatment, direction of symptoms, sputum and extension of the disease. Main results: Patients in DOT and in AT differ in some aspects. In AT, males, older age, positive sputum and more extensive disease were detected, previous treatment predominantly DOT. No difference was detected in relation to duration of symptoms, race, alcoholism and call up. Abandon was more frequent in AT (5.0% vs. 17.0%), p < 0.01, OR 3.90, CI 95% (2.50-5.88); one abandom in AT may be prevented by treating 8.4 patients in DOT. In AT, abandonment was related to male, black race, previous treatment, duration of symptoms, alcoholism and to age; duration of symptoms in DOT. Conclusion and proposal: 1) As abandonment is more common in AT, it is necessary to begin DOT immediately in structured units. 2) Cooperation with company medical facilities, community leaders and pharmacist will make it easier DOT adoption. 3) Punitive measures must be adopted for uncooperative patients.

Keywords: Tuberculosis. Abandonment. Directly-observed treatment (DOT).