Infectious disease from domestic hosts, held for agriculture, can impact wild species that migrate in close proximity, potentially reversing selective advantages afforded by migration. For sockeye salmon in British Columbia, Canada, juveniles migrate past numerous Atlantic salmon farms from which they may acquire a number of infectious agents. We analyse patterns of molecular detection in juvenile sockeye salmon for one bacterial pathogen, Tenacibaculum maritimum, known to cause disease in fish species around the globe and to cause mouthrot disease in farmed Atlantic salmon in BC. Our data show a clear peak in T. maritimum detections in the Discovery Islands region of BC, where sockeye migrate close to salmon farms. Using well established differential-equation models to describe sockeye migration and T. maritimum infection spread, we fit models to our detection data to assess support for multiple hypotheses describing farm- and background-origin infection. Despite a data-constrained inability to resolve certain epidemiological features of the system, such as the relative roles of post-infection mortality and recovery, our models clearly support the role of Discovery-Islands salmon farms in producing the observed patterns. Our best models (with 99.8% empirical model support) describe relatively constant (background) infection pressure, except around Discovery-Islands salmon farms, where farm-origin infection pressure peaked at 12.7 (approximate 95% CI: 4.5 to 31) times background levels. Given the evidence for farm-origin transfer of T. maritimum to Fraser-River sockeye salmon, the severity of associated disease in related species, and the imperilled nature of Fraser River sockeye generally, our results suggest the need for a more precautionary approach to managing farm/wild interactions in sockeye salmon.