SETTING: Vientiane municipality, Laos. OBJECTIVE: To describe and evaluate the implementation of standardized registration and reporting procedures within the municipality in a period of decentralization. The purpose of the standardization was to obtain complete and reliable information on case-finding and treatment results. DESIGN: Evaluation by review of quarterly reports on case-finding and treatment results, and comparison of information from the records of individual patients on site with the records of the supervisors. RESULTS: Case-finding increased in 1994-1998. Treatment success improved from 26.8% (95%CI 21.5-44.0) to 74.6% (95%CI 69.9-79.3). Given an option for auto-administered long-course treatment versus directly observed short-course treatment, 97% of the patients registered in 1998 received the latter. Following decentralization, evaluation of results of treatment was incomplete due to a high ratio of 'transfer out' patients (16.4%, 95%CI 11.5-21.3). After intervention this proportion was reduced to 2.3% (95%CI 0.7-3.9). In 1998, approximately 40% of smear-positive patients diagnosed at the National Tuberculosis Center and residing in the districts evaluated were 'transferred' or 'referred' to the districts. All but three of the 68 patients presented at the district hospitals, mostly without delay. Registration and reporting was reasonably accurate. At the end of the period evaluated, decentralization was only partial and problems in case-holding were still evident. CONCLUSIONS: When decentralizing services it is important to establish procedures for management of information flow in order to allow the activities to be evaluated and problems amenable for correction to be identified.