Abstract:Reeves’s Pheasants (Syrmaticus reevesii) is a threatened species of pheasants, which is endemic to China. The wintering habitat selection by the species was investigated at three scales (10m×10m, 115m and 250m scale, i.e. 0.01 hm2, 4.15 hm2 and 19.63 hm2, correspondingly) in Dongzhai National Natural Reserve from 2000 to 2002. In addition, the characteristics at distance scale were also investigated. At each scale a range of habitat variables were compared between used and control points. At the smallest scale (0.01 hm2), the variables influencing wintering habitat selection were slope, tree coverage, and the interaction between the cosine of slope aspect and shrub height. Reeves’s Pheasant preferred sites were those with an aspect exposed to the sun, a gentle slope, higher shrubs, and high tree coverage. At the mid-scale (4.15 hm2), the area of shrub, broad-leaved forest, and conifer forest were the key factors. Shrub cover was lower around the used points compared to the control points, whereas broadleaf or conifer forest cover was higher. At the largest scale (19.63 hm2), broadleaf and conifer forest cover, and there interaction were the key factors. Habitat use was similar at the mid and large-scale, and it seems that a mosaic of habitats is important to Reeves’ pheasants. Moreover, the proximity of beach and farmland was important of winter habitat selection. According to the lowest AIC and AICc values at the mid-scale, the characteristics at this scale were stated as the ultimate factors influencing the habitat selection of the bird. When considering a range of habitat variables at all scales within a multivariate regression, the most important variables identified at each scale were conifer forest cover at the mid-scale, broadleaf forest cover and the interaction between the conifer forest cover and shrub cover at the large-scale and distance to beach and farmland. Overall there were differences in habitat selection by Reeves’s Pheasant at each scale, and some habitat factors have an interactive effect. These results highlight the important of multi-scale analyses when considering habitat selection by Pheasants.