Examination of processes influencing forage habitat (patch) use is needed to understand factors driving spatial variation in habitat selection (use in relation to availability). I hypothesized that grass abundance was positively associated with use by Roosevelt elk (Cervus elaphus roosevelti) within patches and that elk use of 2 types of patches (clear-cuts and meadows) was supplemental, not complementary. Supplemental use of patches occurs when no single patch provides enough forage (grass in this study) to sustain the population and, hence, grass in one patch substitutes for grass in other patches. Complementary patch use occurs when a resource in one patch is nonsubstitutable with resources in other habitats. Predictions were generated, tested, and upheld for supplemental patch use. Elk use was higher in meadows, which had more grass, than in clear-cuts and elk use of clear-cuts was positively associated with grass biomass. I detected positive relationships among clear-cut size, meadow size, and elk use. Also, elk use of meadows was associated with proximity and size of the next meadow. Spatial variation in patch selection by this population of Roosevelt elk was influenced by grass biomass within patches and attributes among patches. I identified forces influencing patch selection because patch use was supplemental. Consequently, forage attributes within patches were connected to use among patches.