We used location data from radio-collared Canada lynx (Lynx canadensis Kerr, 1792), pellet-count data from snowshoe hares (Lepus americanus Erxleben, 1777), and cover-type data from satellite imagery to evaluate the relationship between the scale of habitat measurement and the potential for persistence of lynx in northeastern Minnesota, USA, at the southern extent of their range. We counted hare pellets at transects throughout northeastern Minnesota to index hare abundance in cover types. Pellet counts were highest in coniferous forest, regenerating-young forest, and shrubby grassland, and these cover types were greater inside lynx use areas than outside of them. Proportions of regenerating-young forest were greater at scales ≥5 km2. We used these results and satellite imagery to map potential lynx core areas. We predicted that 7%-20% of the study area was suitable for lynx. Areas that we predicted to be suitable for lynx corresponded with known core areas, including those withheld from analyses. To maintain habitat for lynx persistence, forest management should retain current levels of 10- to 30-year-old coniferous forest and include ≥5 km2 areas containing 40% of 10- to 30-year-old coniferous forest. Mapping of potential core areas would be improved if cover-type data from satellite imagery identified conifer regeneration.