Despite the demonstrated importance of watershed and regional factors on aquatic ecosystems, inland lake fisheries management is still largely guided by information collected at the local scale.  To better understand how landscape-scale habitat variables influence fisheries resources, we characterized the landscape context of Michigan lakes across multiple spatial scales.  We then used classification trees to develop models predicting presence-absence, abundance, and growth of Michigan lake fishes based upon landscape variables.  Model accuracies ranged from 60 to 80% and included variables that characterize lake position, size, connectivity, and thermal regime.  We then used these models to estimate statewide patterns of species distribution, abundance, and growth.  These estimates can be used to assess ecological condition and develop conservation and management strategies for many lakes across broad spatial extents.