In Great Britain a recent independent scientific review for the government has concluded that the development of a cattle vaccine against Mycobacterium bovis holds the best long-term prospect for tuberculosis control in British herds. A sine qua non for vaccination is the development of a complementary diagnostic test to differentiate between vaccinated animals and those infected with M. bovis so that test-and-slaughter-based control strategies can continue alongside vaccination. In order to assess the feasibility of developing a differential diagnostic test for a live vaccine, we chose M. bovis BCG Pasteur as a model system. Recombinant forms of antigens which are expressed in M. bovis but not, or only at low levels, in BCG Pasteur (ESAT-6, MPB64, MPB70, and MPB83) were produced. These reagents were tested either alone or in combination by using peripheral blood mononuclear cells from M. bovis-infected, BCG-vaccinated, and Mycobacterium avium-sensitized calves. All four antigens induced in vitro proliferation and gamma interferon responses only in M. bovis-infected animals. A cocktail composed of ESAT-6, MPB64, and MPB83 identified infected animals but not those vaccinated with BCG. In addition, promiscuous T-cell epitopes of ESAT- 6, MPB64, and MPB83 were formulated into a peptide cocktail. In T-cell assays with this peptide cocktail, infected animals were identified with frequencies similar to those obtained in assays with the protein cocktail, while BCG- vaccinated or M. avium-sensitized animals did not respond. In summary, our results suggest that peptide and protein cocktails can be designed to discriminate between M. bovis infection and BCG vaccination.