Besides field surveys and experimental studies, modelling is an efficient approach to investigate infectious pathogen spread. BVDV infection is widespread among industrial cattle herds, causing considerable economic losses. Despite control efforts made in European countries, determinants and pathways of BVDV spread between herds, possibly varying with the farming system, were not fully investigated. Our objective was to investigate BVDV spread in a heterogeneous and dynamic bovine metapopulation, accounting for regional specificities. We developed a stochastic multiscale epidemiological model in discrete time; combining a data-driven approach for trade movements (using a comprehensive database) and a mechanistic approach for infection dynamics and demography. Processes were modelled within each of the 19,625 herds located in Brittany (France). Animal trade movements and seasonal neighbouring contacts were considered as between-herd transmission pathways. Demographic rates and the risk of vertical transmission were calibrated on data. Simulated population dynamics were in agreement with observations. We highlighted that neighbouring relationships had a significant periodic effect on the proportion of infected herds over time. Neighbouring effect grows as the radius of the assumed circular neighbouring region increases. Similarly, the probability of virus persistence was higher when considering neighbouring relations. Finally, the within-herd prevalence was impacted by both between-herd transmissions. The epidemiological modelling framework proposed is generically designed and useful for the study of other endemic cattle diseases spread