Borgu (n = 10), Lagoon (n = 10) and White Bororo zebu (n = 10) cattle were experimentally infected in Benin with a clone of Trypanosoma brucei brucei for evaluation of blood antigens using ELISA tests. None of the three non-inoculated control animals for each breed developed any infection. Borgu and Lagoon cattle developed a benign disease followed by spontaneous recovery, whereas White Bororo zebus developed a typical chronic disease, fatal within several months. Overall ELISA sensitivity to detect T. brucei antigens was 20.46%, clearly lower than that of the buffy coat method to detect parasites (40.24%). The sensitivity of ELISA tests differed greatly according to the type of animals and the dynamics of infection. It was highest in Borgu cattle. (44.44 %), which had high parasitaemias of short duration and lowest in White Bororo zebus (4.09%), which had low and intermittent parasitaemias. To increase the sensitivity of ELISA tests by decreasing the cut-off point of optical density (threshold of positivity) from 0.050 to 0.025 would highly compromise the test specificity. The false negatives at the beginning of the infection and false positives after spontaneous cure further compromise the validity of ELISA tests for diagnosis of active T. brucei infections.