Reported are the results of a formal decision analysis which facilitated the choice of the most appropriate test-treatment strategy for visceral leishmaniasis in areas where the disease is endemic. The following strategies were compared: treatment of all suspects (strategy A); testing by means of parasitological investigation followed by treatment of positives (strategy B); two-step testing by means of the direct agglutination test (DAT) followed by treatment of patients with high titres as well as those with parasitologically confirmed borderline titres (strategy C); and DAT followed by treatment of positives (strategy D). The results for each strategy were expressed as costs in US$ per death averted. The effectiveness of strategies C and D was close to that of strategy A and far better than that of strategy B. The cost-effectiveness ratio for strategies C and D (US$ 465 per death averted) was not substantially higher than that of testing by means of parasitological investigation followed by treatment of positives (strategy B), which was the most cost-effective strategy at US$448 per death averted. At current prices of antimonial drugs, the cost of test-treatment strategies depends more on the cost of treatment than on that of testing. The use of a sensitive serological test such as the DAT is recommended as the basis of test-treatment strategies for visceral leishmaniasis in areas where the disease is endemic.