A Johne's disease control program based on herd testing with a commercial enzyme-linked immunosobent assay (ELISA) and fecal culture, was implemented in 25 Victorian dairy herds. A whole herd screening test using only the ELISA was conducted at the beginning of the project in 1989-1990 to estimate disease prevalence in each herd, and to allow the progression of disease control in each herd to be modeled using the Collins-Morgan model of Johne's disease epidemiology. Subsequently, while implementing recognized management strategies to control the disease, animals were tested annually with the ELISA and fecal culture. Test positive animals were removed annually prior to the calving period. At the end of the study, the prevalence of disease had decreased from an average herd prevalence of 9.7% in 1990 to 3.2% in June 1993. There were 41 cases of clinical Johne's disease in the study herds in the period July 1990 to June 1991, but only 3 cases of clinical Johne's disease reported from herds culling test positive animals, in the period July 1992 to June 1993. Johne's disease was confirmed by histopathology or culture of post-mortem specimens from 86% of 282 ELISA or fecal culture positive animals examined. Mycobacterium paratuberculosis was cultured from 8 (9.2%) of fetuses from 87 subclinically infected ELISA positive cattle, and 6 (50%) of 12 clinically affected ELISA positive cattle. When testing was conducted more than once during each year the greatest number of test positive animals were detected at the mid lactation test. In two herds, no difference was detected between the average production index for ELISA positive cattle and the average production index for the rest of the herd. The results from this study indicate that the Collins-Morgan model can be used to broadly anticipate the results of Johne's disease control in Victorian dairy herds, at least in the early stages of the program. At the end of the project none of the herds could be considered to be free of M. paratuberculosis infection. Nevertheless, randomly selected cattle from these herds are considerably less likely to be M. paratuberculosis infected than similar cattle selected from the general Victorian dairy cattle population. Recommendations are made regarding the implementation of Johne's disease control programs in endemically infected dairy herds.