To assess the productivity and cost effectiveness of a traditional treatment monitoring protocol, the records of all 268 patients entered into the tuberculosis treatment program of Denver's municipal clinic during 1968 to 1972 inclusive were analyzed. Treatment failures were detected in 7 patients undergoing therapy, and disease reactivation was noted in 2 patients after treatment. Factors leading to the recognition of these failures were evaluated. It was observed that routine sputum cultures obtained only from selected, higher-risk patients would have detected all but one of the treatment failures occurring during therapy, whereas symptomatology would have identified both post-therapy reactivations. The cost effectiveness data presented suggest that substantial modifications of the methods for monitoring the outcome of tuberculosis chemotherapy should be made.