The objective of this research was to classify Darkola oriental beech (Fagus orientalis Lipsky) plant communities in Pol-sefid in Mazandaran Province on the basis of the above ground vegetation and soil seed bank floristic dataset. We also tested whether above ground plant communities can be recognized by a soil seed bank dataset. For this purpose 52 releves with an area 400 m2 were made at the peak of the growing season during June 2006 by a systematic-selective method and by consideration of the indicator stands concept. Vegetation data were recorded separately for each life form. In each releve, soil samples were also collected using 20 cm × 20 cm square metal frame in six repetitions at the beginning of the 2007 growth season. The metal frame was hammered into the soil to a depth 10 cm. This study used the seedling emergence approach to recognize the size and richness of species composition in the seed bank. By using two way indicator species analysis, TWINSPAN, four plant communities were separately recognized in two series as above ground plant communities and underground (soil seed bank) plant communities based on the relative density of total plant species identified in the soil seed bank and the percentage cover of each species in above ground vegetation dataset matrices surveyed, respectively. Above ground and underground plant communities’ fitness was estimated at 68% based on similar membership of releves in each plant community, when they were separately classified from the soil seed bank and above ground vegetation points of view. Results of discriminant analysis and detrended analysis revealed that classification of plant communities on the basis of the above ground vegetation dataset led to forming distinguishable and separate groups whereas the soil seed bank dataset, since it was generally composed of pioneer plant species, could not display distinct plant communities. As a result, we concluded that soil seed bank floristic data were not suitable for plant community classification even though they were related to some physiographical properties. The results of indicator species analysis (IV) approved that there are four groups of indicator plants which characterized Darkola oriental beech forest into four distanced above ground plant communities including: Fagus orientalis with Danae racemosaunderstory type, Fagus orientalis- Acer velutinum types, Fagus orientalis with Mercurialis perennis understory and Fagus orientalis with Vaccinium arctostaphylos understory.