This study was carried out to develop the regression model for estimating biomass of natural Pinus densiflora forests by stand density in northeast Chinese area of Mt. Paekdu. Four allometric regression models(W=aD$^b$, W=a(D$^2$H)$^b$. logW=a+b$cdot$ logD+cD and logW=a+b$cdot$log(D$^2$H)+c(D$^2$H)) were used to estimate biomass for each of the tree components. The suitable regression model for estimating biomass of stem, bark and whole tree above ground was logW=a+b$cdot$log(D$^2$H)+c(D$^2$H), and that for biomass of branch, needle and needle area, logW=a+b$cdot$logD+cD for all of the stand density classes.