In this study, 50 common plant species in the karst rocky mountains of Guilin were selected, and contents of carbon (C), nitrogen (N), phosphorus (P), potassium (K) and calcium (Ca) in leaves were measured. Multivariate analysis was adopted to characterize the nutrient compositions of the selected plants in order to explore their adaptability to the habitat of karst rocky mountains aiming to provide guidance for vegetation restoration in the karst areas. Variance analyses showed that there were significant differences among the selected plant species in terms of the indices of nutrient contents, including C_mass, N_mass, P_mass, K_mass, Ca_mass, C/N, C/P and N/P in the leaves. The contents were highest for C and Ca, intermediate for N and P, indicating that growth of these plant species were mainly limited by P. The correlation analyses revealed that C% was extremely significantly (P < 0.01) correlated with N% and C/N, and was significantly (P < 0.05) correlated with K% and Ca%. N% was extremely significantly correlated with C/N, C/P, P% and K%. P% was extremely significantly correlated with K%, C/N and N/P, and was significantly correlated with C/P. The correlation between K% and C/N was extremely significant. C/N, N/P and C/P were extremely significantly correlated. The correlations of other indices were not significant. Cluster analyses based on leaf nutrients indicated that the selected plant species could be divided into three different adaptive types, i.e., calcium-loving plants, highly productive plants, and drought-resistant plants. Our results provide useful implications for the reconstruction and restoration of vegetation in the karst areas.