The conservation of wide-ranging species presents challenges in a world of increasingly human-dominated landscapes, which forces animals to occupy human-modified landscapes. Although analyzing habitat selection and identifying suitable habitats are important in supporting natural recolonization, these actions are rarely validated owing to difficulties in monitoring animal colonization events. Habitat studies are, however, fundamentally important for species conservation policies and management decisions. Analyses of habitat selection and evaluation for animal conservation, especially moving and bedding behaviors, are often in short supply. We conducted a study on habitat selection and evaluation in roe deer (Capreolus pygargus), which is one of the main prey species of the Amur tiger (Panthera tigris altaica) in southern Laoyeling, Heilongjiang Province. A large part of southern Laoyeling is inhabited by the Amur tiger and the Leopard National Park and is a key habitat for tigers and ungulates. It is one of the main distribution regions of the Amur tiger and its prey. Systematic studies are required to thoroughly investigate the effects of landscape factors and tiger occurrences on the selection of suitable habitats by roe deer, which are necessary for wildlife conservation. Such data were collected by conducting 28 plot surveys and 84 lines for footprint chain tracking during the winters from 2017 to 2019 in southern Laoyeling of Heilongjiang Province. Furthermore, data of tiger occurrences were recently collected and integrated with roe deer data into a habitat analysis. Generalized additive model (GAM) and Maximum Entropy (MaxEnt), based on ecological niche theory, were used to assess the response of roe deer to different landscape factors and to evaluate the suitability of habitats for roe deer during the snow season. From our investigation of roe deer movement between habitats, we found that roe deer preferred habitats including flat with small slopes, less than 500 m from farmland, at low or high-elevation areas, and in areas without roads and settlements. The results of our habitat evaluation for movement without tiger occurrence data showed that the combined area of suitable large and medium-sized habitats was 1318.16 km² (51.28% of the study area), but was only 901.52 km² (a decline of 31.61%) if tiger occurrence data were used as a factor in the habitat selection model for roe deer. For bedding habitats, our results demonstrate that factors including river, farmland, road and snow depth affect the habitat selection of roe deer. Among them, snow depth contributed 70.13% to the bedding habitat selection of roe deer. The results of bedding habitat evaluation, without tiger occurrence data, showed that the combined area of suitable large and medium-sized habitats was 1243.77 km² (48.39% of the study area), but was only 485.02 km² (decline of 61.00%) if tiger occurrence data were included as a factor in the roe deer habitat selection model. Our results suggest that predator-prey interactions may result in strong spatial patterns of avoidance. We suggest that the occurrence of tigers has an ecological effect on the habitat selection of roe deer and makes habitats less suitable for roe deer. Roe deer may, in fact, be more likely to avoid specific habitats or landscape features that increase their vulnerability to predation.