Proper forest management needs quantitative and precise estimates of forest stands characteristics. Remotely sensed imageries, due to accurate and broad spatial information, has become a cost-effective tool in forest management. Classification of forest attributes and generation of thematic maps are among the common applications of remote sensing. The objective of this study was to optimize Random Forest algorithm for classification of quantitative attributes of Manesht forest in Ilam Province. Two parameters including mtry= 8, 8, 6 and ntree =300, 800, 200 were used as the optimum numbers to classify basal area, canopy cover and density, respectively. The results showed the more accurate classification in canopy cover (overall accuracy=83%, Kappa coefficient=0. 73), basal area (overall accuracy=78%, Kappa coefficient=0. 72) and density (overall accuracy=75%, Kappa coefficient=0. 69), respectively. Furthermore, variable importance index indicated distance-based vegetation indices are more important for basal area and density classification. It is concluded that the Random Forest algorithm as a non-parametric method could classify basal area, canopy cover and density properly.