Abstract In the ruling to list Canada lynx ( Lynx canadensis ) as a federally threatened species, the U.S. Fish and Wildlife Service (USFWS) identified the Great Lakes region as an area that historically contained lynx and, hence, could potentially contribute to population recovery. More recent critical habitat designations by the USFWS only recognize Minnesota, USA as important to recovery in the Great Lakes. Although there is no current evidence of a resident lynx population in the Upper Peninsula (UP) of Michigan, USA, trapping and track records over the past century suggest the region was periodically invaded after lynx population irruptions in Canada. In support of state and federal agency efforts in Michigan to provide and conserve lynx habitat, we quantified habitat potential using a spatially explicit, landscape‐level model based on relationships among lynx, their primary prey, snowshoe hare ( Lepus americanus ), and vegetation attributes. Outputs from the model indicated that habitat in the UP supports low hare densities (<0.07–0.75 hares/ha). Corresponding potential lynx densities ranged from 0/100 km 2 in the southern and northeast UP to 5/100 km 2 in the central–eastern UP. Model estimates of potential hare density were correlated with winter track‐surveys ( R 2 = 0.4, P < 0.001). Current absence of a resident lynx population in Michigan is likely attributed to confounding factors (e.g., habitat, competition, status of source population) but our results indicate that current habitat quality, quantity, and spatial configuration are exerting large‐scale negative influences. These results are generally consistent with the USFWS determination that Michigan's UP most likely functions as dispersal habitat. © 2011 The Wildlife Society.