Greater sage-grouse (Centrocercus urophasianus) at the western edge of the Dakotas occur in the transition zone between sagebrush and grassland communities. These mixed sagebrush (Artemisia sp.) and grasslands differ from those habitats that comprise the central portions of the sage-grouse range; yet, no information is available on winter habitat selection within this region of their distribution. We evaluated factors influencing greater sage-grouse winter habitat use in North Dakota during 2005–2006 and 2006–2007 and in South Dakota during 2006–2007 and 2007–2008. We captured and radio-marked 97 breeding-age females and 54 breeding-age males from 2005 to 2007 and quantified habitat selection for 98 of these birds that were alive during winter. We collected habitat measurements at 340 (177 ND, 163 SD) sage-grouse use sites and 680 random (340 each at 250 m and 500 m from locations) dependent sites. Use sites differed from random sites with greater percent sagebrush cover (14.75% use vs. 7.29% random; P < 0.001), percent total vegetation cover (36.76% use vs. 32.96% random; P ≤ 0.001), and sagebrush density (2.12 plants/m² use vs. 0.94 plants/m² random; P ≤ 0.001), but lesser percent grass cover (11.76% use vs. 16.01% random; P ≤ 0.001) and litter cover (4.34% use vs. 5.55% random; P = 0.001) and lower sagebrush height (20.02 cm use vs. 21.35 cm random; P = 0.13) and grass height (21.47 cm use vs. 23.21 cm random; P = 0.15). We used conditional logistic regression to estimate winter habitat selection by sage-grouse on continuous scales. The model sagebrush cover + sagebrush height + sagebrush cover × sagebrush height (w_i = 0.60) was the most supported of the 13 models we considered, indicating that percent sagebrush cover strongly influenced selection. Logistic odds ratios indicated that the probability of selection by sage-grouse increased by 1.867 for every 1% increase in sagebrush cover (95% CI = 1.627–2.141) and by 1.041 for every 1 cm increase in sagebrush height (95% CI = 1.002–1.082). The interaction between percent sagebrush canopy cover and sagebrush height (β = −0.01, SE ≤ 0.01; odds ratio = 0.987 [95% CI = 0.983–0.992]) also was significant. Management could focus on avoiding additional loss of sagebrush habitat, identifying areas of critical winter habitat, and implementing management actions based on causal mechanisms (e.g., soil moisture, precipitation) that affect sagebrush community structure in this region.