Farmed non-native Atlantic salmon (Salmo salar) is the largest agriculture export product of British Columbia, Canada. Chronic low-volume escapes of salmon from farms into Pacific waters (“leakage”) are typically undetectable (Britton et al. 2011). Analysis of escape-reporting from farmers indicates that reports greatly underestimate the true number of Atlantic salmon inadvertently released from open-net pen rearing sites (Morton and Volpe 2002). To quantify the spatial extent of escaped Atlantic salmon in Canadian Pacific rivers, we systematically snorkel-surveyed 41 known Pacific salmon (Oncorhynchus spp.)-supporting rivers and creeks on Vancouver Island over a span of 3 years. We estimated and accounted for imperfect detections using multi-season occupancy models. We detected Atlantic salmon in 36.6 % of surveyed rivers. After accounting for imperfect detection, occupancy models estimated that over half of surveyed streams across the study area contained Atlantic salmon, and that 97 % of streams with high native salmon diversity were occupied by Atlantic salmon. Even in intensive snorkel surveys, Atlantic salmon are detected in occupied streams only 2/3 the time, suggesting abundance and distribution of non-native salmon is greater than indicated by the only existing data. Further, Atlantic salmon are more likely to occupy streams with high native Pacific salmon diversity—and more likely to maintain occupancy across years—potentially increasing competitive pressure on native salmonids. Understanding local biotic and abiotic predictors of Atlantic salmon occupancy, stream colonization, and local extinction requires more data; the same is true for the effects of escaped Atlantic salmon on local salmon diversity and sustainability. These data for the first time show that Atlantic salmon occupy Pacific coastal rivers for multiple years. The impact of Atlantic salmon occupancy in British Columbia rivers must be factored into policy decisions regarding the future of salmon farming in the provincial waters.