SETTING: National Tuberculosis Programme (NTP), Cotonou, Benin. OBJECTIVE: To study the patient characteristics and outcome of tuberculosis retreatment cases in a well-functioning NTP. METHODS: A retrospective, register-based study of all smear-positive pulmonary tuberculosis cases put on retreatment (2SERHZ/1ERHZ/5R3H3E3) between 1992 and 2001 in Cotonou. For comparison, information on new smear-positive cases in Cotonou in 1999 was entered and analysed. RESULTS: Of 8103 tuberculosis patients registered, 642 were put on retreatment. The analysis is mainly based on the 236 patients whose initial treatment regimen records were available (113 relapses, 84 failures, 39 returns after default). Most of the relapse (57%) and return after default (72%) cases were put on retreatment within 12 months after stopping their initial treatment. Overall, the retreatment results were satisfactory (78% success) and comparable with those of new cases (82%); the failure rates were low (3%), as were those for initial treatment (1%). There were more defaulters from retreatment among those who had already defaulted from initial treatment (21%). Treatment success rates were better among women than men. CONCLUSION: The standardised retreatment regimen is effective in Cotonou, probably because the NTP is functioning well, there are no drug shortages, drug taking is strictly supervised, and a good treatment plan is followed.