Control of endemic visceral leishmaniasis (VL) in large communities requires a feasible epidemiological indicator capable of monitoring on-going transmission rather than mere exposure to the parasite. Following confirmation of the desired level of reliability for laboratory diagnosis of VL, the direct agglutination test (DAT) was employed to estimate VL sero-prevalence in the endemic upazilas (subdistricts) of Trishal and Shahjadpur within Mymensingh and Sirajganj districts of Bangladesh. DAT antigen production was duly increased to allow coverage of a study population of 17826 inhabitants, 9619 of whom resided in Trishal, 7328 in Shahjadpur and 879 in Teknaf (Cox's Bazar), a known Leishmania-free district in Bangladesh. Despite large-scale production in batches of 1120-4000 ml (each sufficient for 1176-6400 screening doses), all DAT antigen batches performed as required in quality control tests for sensitivity, specificity and stability. It was convenient for both collection and testing to take the required samples of whole blood by finger prick. A cross-sectional survey revealed VL point prevalences of 4.40% in Trishal and 6.75% in Shahjadpur, compared with an extremely low rate of 0.34% in non-endemic Teknaf. In both endemic upazilas (Trishal and Shahjadpur) VL was more prevalent (2.56-4.5%) in persons up to 20 years of age than in those 21 years of age and older (1.84-2.25%). Of 918 subjects recorded as seropositives, 539 were VL-asymptomatic and 379 were VL-symptomatic with various degrees of suspicion. Diagnosis of VL was established in 125 symptomatic seropositives subjects, either by demonstrating the presence of Leishmania amastigotes (29), or by positive DAT results combined with presentation of typical VL signs (96).(ABSTRACT TRUNCATED AT 250 WORDS)