Johne’s disease, a severe granulomatous enteritis of ruminant animals, is caused by Mycobacterium avium subsp. paratuberculosis (MAP). MAP infections have detrimental consequences for animal health and reduce dairy-herd productivity. Bacterial fastidiousness and slow generation time encumber diagnostic testing strategies. MAP is also a potential etiologic agent of human Crohn’s disease, with the bulk-milk supply serving as a possible transmission vector. The objective of this dissertation is to explore MAP infection dynamics on dairy farms, with an emphasis on the routes of bulk-milk contamination, transmission risk across production type, and the interplay between diagnostic testing outcomes. Accordingly, we have applied statistical and mathematical approaches to both cross-sectional and longitudinal datasets. Using questionnaire data from 292 U.S. dairies, we conducted a comparative risk assessment of organic vs. conventional management and determined that organic herds were at higher risk for new MAP infections. We concluded, empirically, that organic farms were more susceptible to a synergism of risk factors within the maternity pen and should improve calving-area hygiene if electing to permit cow-calf contact. Bulk-milk testing was also conducted for these herds. Most high ELISA tanks were PCR negative, implying that ELISA is not a perfect predictor of bulk-milk MAP status; for accurate risk assessment, bulk-milk ELISA should be used in tandem with PCR. A combination of ELISA and PCR may also aid in determining the specific route of bulk-milk contamination (either environmental or direct shedding). To extend the investigation to individual animals, longitudinal data were obtained from 14 MAP-positive cows in 2 low-prevalence herds. Robust relationships between culture, fecal qPCR, and milk ELISA were revealed, using mixed linear modeling to adjust for cow characteristics. We explored temporal relationships and observed that spikes in fecal shedding were predictive of subsequent high milk ELISA results. We also noted that disease “Progressors,” (infected animals with increasing fecal MAP CFU over time) had higher antibody titers overall. Interestingly, the paucity of positive milk samples, from both individual and bulk- tank sources, suggests that milk contamination is not a chief concern in low-prevalence herds. Armed with insights from these studies, in addition to published literature, we developed a mathematical model to explore the interaction between categories of infection, environmental MAP burden, and bulk-tank contamination. Direct shedding into milk accounted for < 1% of the MAP CFU in the tank, with environmental contamination from high shedders as the primary driver of bulk-milk MAP burden. Culling of high shedders, cleaning of the maternity pen, and adherence to milking parlor cleanliness each had a strong influence on lowering the bulk-milk MAP load. A combination of these initiatives served to drive the MAP level below an acceptable threshold (< 103 CFU/L). While complete elimination of MAP may be an unrealistic target for high-prevalence herds, the production of bulk milk with a low MAP load appears feasible. In this work, we assess the significance of a variety of contamination routes, transmission risks, and intervention strategies. These efforts are directed toward improved understanding of testing schemes and an ultimate refinement of control measures and milk quality programs. The conclusions from the studies presented in this dissertation may be applied to mitigate the spread of MAP in dairy herds, reduce prevalence, and lower or eliminate MAP in the bulk-milk supply.