Methods used to diagnose Trypanosoma cruzi infection differ in their ability to discriminate between sera from infected and uninfected individuals. We compared the results of an immunofluorescence (IF) test, a haemagglutination (HA) test, and an enzyme-linked immunosorbent assay (ELISA) for the diagnosis of T. cruzi infections in a large population-based survey in central Brazil using blood eluates from filter-paper and venous blood samples. The sensitivities of the tests on eluates, compared with results on serum samples, were low: ELISA (78.1%), IF (69.2%) and HA (64.6%). The level of agreement between the tests on eluates was very poor, with the best co-positivity for IF and ELISA. Both the positive and negative predictive values of the three tests on eluates were similar (around 96%) to those for sera. Higher co-positivity values were obtained for the three tests on sera. The implications of these results are discussed in relation to blood screening, routine medical practice, sero-epidemiological surveys, and the follow-up of patients admitted to therapeutic trials.