132 patients with suspected kala-azar (visceral leishmaniasis) were included in a prospective study to compare the performance of the direct agglutination test (DAT) with smears of lymph node, bone marrow and splenic aspiration. A titre greater than 1:3200 was considered positive. 67 patients provided positive smears and 65 were smear-negative. Compared with the results obtained from smears, the sensitivity of the DAT was 94% and its specificity 72%. Of 18 patients who were negative by smears but positive by DAT, 3 were treated on strong clinical suspicion of kala-azar and responded well to therapy. One other patient had post-kala-azar dermal leishmaniasis. Of the remaining 14, 8 were tested with the leishmanin skin test; 6 gave a positive result, suggesting past or sub-clinical infection. In 4 smear-positive patients, the DAT was negative; in 2 of these, the test remained negative during 6 months' follow-up. In a sero-epidemiological survey in a camp for displaced people, 30 of 600 sampled individuals gave a positive DAT. Of these, 11 had been treated for kala-azar; 4 others were leishmanin positive. The DAT is a useful screening test (sensitivity 94%; predictive value of a negative test 92%), but it does not differentiate between past kala-azar, sub-clinical infection and active disease.