The rate of spread of bovine brucellosis was investigated in buffalo and Jersey cattle herds maintained at a Livestock Research station in Pakistan. Reactor animals (identified by conventional serological tests) were either retained or culled because of advanced age or poor productivity. Reactors were housed, managed and fed separately from the rest of the herds. The initial seroprevalence of brucellosis among both the Jersey cattle and the buffalo tested was 21.4%, the difference being statistically insignificant (p = 0.218). For 34 months, the spread of brucellosis was limited to 25 new reactors in the 334 cows and 33 in the 442 buffaloes. The mean attack rate was 7.5% for both herds during the test intervals. Trend analysis of proportions positive at each testing revealed a significant decrease in the percentages observed at the first testing. The management practice of segregation offered some advantage in reducing the spread of brucellosis to negative animals. However, an epidemiological study covering a large number of herds would be required to identify risk factors responsible for perpetuating the disease.