An enzyme-linked immunosorbent assay (ELISA) for the diagnosis of bovine tuberculosis through the detection of specific seric antibodies has recently been developed in our laboratory. In order to assess its reproducibility and select the most adequate antigen, four bovine PPDs from different sources were evaluated in parallel: PPD M. bovis strain AN5, CEPANZO standard (CPZ), PPD M. bovis strain AN5, European Economic Community standard (EEC), PPD M. bovis strain AN5, prepared from non heated bacilli, killed by phenol (P) and PPD. M. bovis BCG strain prepared at the Pasteur Institute, Paris (BCG). Sera from 22 healthy cattle from tuberculosis free area and 20 bacteriologically confirmed tuberculous animals were employed in simultaneous assays. Antibody mean and standard deviations from healthy cattle expressed as optical density (OD) values were 45 +/- 22 when CPZ was used as antigen, 24 +/- 10 with EEC, 103 +/- 56 with P and 56 +/- 20 with BCG. Mean O.D. from tuberculous cattle were 588 +/- 158, 510 +/- 234, 782 +/- 138 and 441 +/- 189 with antigens CPZ, EEC, P and BCG respectively. A close correlation was observed when results obtained with EEC and P were compared with that of CPZ (r: 0.97 and 0.94 respectively). A lower specificity was achieved when BCG was used as antigen being also lower its correlation with the results obtained with CPZ (r: 0.87). It is concluded that our ELISA would achieve similar sensitivity and specificity if CPZ, EEC and P were used as antigens. On the other hand, BCG would not be suitable for this assay.