High-mountain coniferous forests widely distribute in varies topography above 2500 m a.s.l in Taiwan, while major dominant coniferous species of high-mountain coniferous forests are Abies kawakamii, Juniperus squamata and Tsuga chinensis var. formosana. Since the high-mountain plant communities are vulnerable to climate change, we selected plots dominated by Abies kawakamii to find potential influential environmental variables and their relations to vegetation composition. Species data were from National Vegetation Diversity Inventory and Mapping Project in Taiwan and distinguished vegetation types of Abies kawakamii communities by ordination analysis. In order to confirm the strength of linkage among environmental variables and vegetation composition, structural equation model (SEM) was used to quantify the coefficients of pathways. Environmental variables were defined as latent variables, such as climate and habitat substrate. Ordination result shows elevation, soil rockiness, rock ratio, warmth index and precipitation in winter play important roles in vegetation composition of Abies kawakamii communities. SEM result indicates climate has higher coefficients than habitat substrate.