Going beyond farm areas, enzootic diseases spread in livestock populations at a regional scale through animal movements between holdings and neighboring relationships. Because animal trade networks are time varying, it is important to be studied over several years to identify if structural patterns underlying pathogen spread are stable in time. In addition, at the intra-herd scale, potentially large heterogeneities occur in the infection prevalence over time (determined by the time since infection and the rate of external exposure), and among herds in a region (with respect to herd specificities such as size and degree of connectivity). Therefore, the intra-herd infection dynamics may have to be accounted for when modelling the risk of pathogen spread to in-contact herds. We first analyzed the French cattle movement network from 2005 to 2009 using tools from graph theory, for different spatial granularities (farms or communes as nodes) and temporal windows (month, year) to investigate the stability of its main descriptors. Proxies for pathogen spread (at the national or regional scales, on the entire network or on subsets of it), such as percolation and reachability ratio, accounting for its time-varying properties, were also computed. Second, we coupled intra- and inter-herd infection dynamics, described by stochastic models in discrete time, to model the spread of bovine viral diarrhea virus (BVDV), as an example of bovine pathogen, on the sub-network of dairy herds in Brittany (France), a main region for dairy cattle farming systems. The location of virus introduction in the region as well as herd specificities highly influenced BVDV spread. The relative roles in inter-herd transmission of neighborhood relationships (occurring in high-density area with long lasting pasture periods) and animal trade were also investigated through sensitivity analysis. Third, we evaluated control strategies related to the indirect knowledge of the source herd infection status, such as the level of antibodies in the bull tank milk, or the imperfect and delayed knowledge of the animal infection status at movement occurrence. The model is currently under extension to account for the heterogeneity in herd farming systems as encountered in a region.