This study was carried out to classify forest types by using multivariate statistical analysis method and to evaluate the ecological attributes of the types in the natural forest of Gariwangsan(Mt.) area . The vegetation data were collected from 183 temporary sample plots by square plot sampling (20 m × 20 m) method. Based on the surveyed vegetation data, cluster analysis and multiple discriminant analysis (MDA) were used to make disordered forests into several common groups. Six forest types were identified through cluster analysis and 93.4% were appropriately classified by multiple discriminant analysis. Recognized forest types were 1) Quercus mongolica forest type, 2) Q. mongolica?Fraxinus rhynchophylla forest type, 3) Betula schmidtii?Q. mongolica?Pinus densiflora forest type, 4) Mixed mesophytic forest type, 5) P. densiflora?Q. mongolica forest type, 6) B. constata?Ulmus laciniata forest type. The species composition, species diversity and other environmental factors of six forest types were analyzed and their ecological characteristics were identified. The proportion of B. schmidtii, P. densiflora, and B. costata which had the highest dominance in the overstory, was decreased and the number of other species in the midstory would be increased, mainly due to their shade?tolerance. The Q. mongolica forest type was most widely distributed over the study area, and the species was located in the ridge with poor moisture condition and showed the lowest species diversity index. Mixed mesophytic forest type was mainly distributed between valleys and mid-slopes with good moisture condition. The largest number of species were presented in the type and showed the highest species diversity index. The natural forests with complex terrain showed different characteristics depending on the microtophography. The species diversity was higher in the valley than in ridge and higher in the northern slopes than in the south. Compared with the ridge, the valley tended to have better moisture conditions.