Control strategies of bovine paratuberculosis at herd scale are currently based on hygiene improvement and test-and-cull of infected animals. These actions are difficult to implement by farmers and lack effectiveness to reduce infection prevalence. Genetic selection could enhance disease control if focused on phenotypic traits of animals which influence the disease dynamics in a population. However, there is a lack of knowledge on which traits contribute the most to the disease dynamics.Our objective was to assess which phenotypic traits should be targeted to enhance paratuberculosis control in dairy cattle herds.We used a within-herd transmission model where we modified traits related to the course of infection dynamics in an animal. The model is stochastic and simulates herd population dynamics and infection course in animals. It accounts for all known transmission routes, shedding heterogeneity, and age related susceptibility. We studied model outcomes describing the disease dynamics in the herd: infection persistence, prevalence, proportion of affected animals, and cumulated number of newly infected animals, 25 years after infection first introduction. We compared a situation assuming phenotypic traits at their current values with situations where animals more resistant to infection would have been successfully selected. Each scenario was defined as a combination of improved phenotypic traits (i.e. reduced susceptibility and reduced duration of the susceptible state, shedding level and durations in infection stages). Cluster analyses were performed to highlight which traits must be improved and to what extent on, to contain the disease dynamics in the herd.Phenotypes combining a high reduction of susceptibility (e.g. halved) with short susceptibility duration (e.g. halved) were the most effective, lowering the probability of infection persistence in the herd and the prevalence in infected herds (e.g. decreased of 60% and 75%, respectively). If, in addition, the duration before entering the affected stage was increased (e.g. doubled) and the shedding level of affected animals was decreased (e.g. halved), the probability of persistence and the prevalence further decreased (e.g. by 92% and 99%, respectively). Modifying other traits was much less effective.Gaining knowledge on phenotypic traits of resistance to be targeted is useful to orientate future efforts in genetics. Regarding the long delays of genetic selection, we should evaluate the potential effectiveness and time needed to control paratuberculosis when combining it with current control measures, starting from an enzootic situation. This study was funded by INRA (métaprogramme GISA), Apis-Gene, and GDS France.