We studied feeding intake and food selection of nine captive forest musk deer (Moschus berezovskii) offered 17 species of plants in China. We also determined nutrient characteristics related to plant quality to assess their effect on food selection. Results indicated that forest musk deer exhibited positive selectivity for four species of plants (M. azedarach, M. baccata, K. japonica and C. orbiculatus) and negative selectivity for the remainder. Two plant species with the highest selectivity values accounted for 47.39 % of total food intake; thus, forest musk deer exhibited the strongest preference for these species. Food intake was positively correlated with feeding frequency and duration (r = 0.764, p < 0.005; r = 0.843, p < 0.005) but was not correlated with sniffing frequency or duration. However, olfaction did play an important role in food recognition by the deer. Pearson correlation analysis (data were log10 transformed) indicated that leaf intake was positively correlated with crude protein content (r = 0.708, p = 0.001) and negatively correlated with crude fiber content (r = –0.811, p < 0.001) and ash content (r = –0.496, p = 0.043). In addition, forest musk deer preferred tannin-rich plants with high protein and low fiber. Food intake was also positively correlated with potassium content (r = 0.672, p < 0.005). Our results suggest that forest musk deer is able to positively select high quality food (high protein content) and avoid low quality food (high fiber content). However, the fact that musk deer also prefer tannin-rich food requires further research to gain deeper insight into the underlying mechanisms in the food selection of forest musk deer.