Protection against Brucella abortus induced abortion and infection provided by strain 19 (S19) vaccination was evaluated in American bison (Bison bison). Forty-eight pregnant bison were manually inoculated (MI) with S19 vaccine, 44 were ballistically inoculated (BI) with an absorbable hollow pellet containing lyophilized S19, and 46 were manually injected with buffered saline as non-vaccinated controls (NVC). All bison were Brucella spp. seronegative prior to the experiment, in the second trimester of pregnancy, and were randomly assigned to experimental groups. Approximately 60 days post-vaccination, abortions were observed in the vaccinated bison. Brucella abortus strain 19 was recovered from a bison that had recently aborted, her fetus, and from 11 of 12 other aborted fetuses. Fifty-eight percent (53 of 92) of vaccinated bison aborted, and no abortions were observed in the NVC bison. One cow aborted during her second post-vaccinal pregnancy and S19 was identified from the dam and fetus indicating that chronic S19 infections can occur in bison. Positive antibody titers were present 10 mo post-vaccination in 73% (66 of 91) of the bison. Thirteen mo post-vaccination, 30 MI vaccinates, 27 BI vaccinates, and 30 NVC bison were challenged during the second trimester of pregnancy with 1 x 10(7) CFU of B. abortus strain 2308 via bilateral conjunctival inoculation. Protection against abortion was 67% (P less than or equal to 0.0001) for vaccinated bison compared to 4% in NVC. Protection against B. abortus infection was determined to be 39% (P greater than or equal to 0.001) for vaccinates and 0% (zero of 30) for NCV. Persistent antibody titers, vaccine induced abortions, and chronic S19 infections indicate that the S19 vaccine doses used in this study are not suitable for pregnant bison.