Predicting potential habitat across a landscape for rare species is extremely challenging. However, partitioned Mahalanobis D2 methods avoid pitfalls commonly encountered when surveying rare species by using data collected only at known species locations. Minimum habitat requirements are then determined by examining a principal components analysis to find consistent habitat characteristics across known locations. We used partitioned D 2 methods to examine minimum habitat requirements of Indiana bats (Myotis sodalis) in the Champlain Valley of Vermont and New York, USA, across 7 spatial scales and map potential habitat for the species throughout the same area. We radiotracked 24 female Indiana bats to their roost trees and across their nighttime foraging areas to collect habitat characteristics at 7 spatial scales: 1) roost trees, 2) 0.1-ha circular plots surrounding the roost trees, 3) home ranges, and 4-7) 0.5-km, 1-km, 2-km, and 3-km buffers surrounding the roost tree. Roost trees (n = 50) typically were tall, dead, large-diameter trees with exfoliating bark, located at low elevations and close to water. Trees surrounding roosts typically were smaller in diameter and shorter in height, but they had greater soundness than the roost trees. We documented 14 home ranges in areas of diverse, patchy land cover types that were close to water with east-facing aspects. Across all landscape extents, area of forest within roost-tree buffers and the aspect across those buffers were the most consistent features. Predictive maps indicated that suitable habitat ranged from 4.7-8.1% of the area examined within the Champlain Valley. These habitat models further understanding of Indiana bat summer habitat by indicating minimum habitat characteristics at multiple scales and can be used to aid management decisions by highlighting potential habitat. Nonetheless, information on juvenile production and recruitment is lacking; therefore, assessments of Indiana bat habitat quality in the region are still incomplete.