The standardization of ELISA for the detection of anti-Trypanosoma evansi antibodies in naturally and experimentally infected horses is described. Bayesian analysis was used to establish the cutoff between positive and negative sera. In order to determine the assessment of the ELISA test, the results obtained were compared with those from an IFA. A relative sensibility of 98.39%, a specificity of 95.12% and a predictive value of 96.83% were determined. The standardized technique was used to evaluate the antibody production against trypanosome in an experimentally infected equine, in which the sera converted 15 days after infection. The test was also used for a study of sera prevalence in a non-random sample from two different populations. A prevalence of 81.7% in workhorse and 57.14% in stable horses was found.