To investigate the effect of size and density on the segregation of particles in the static classifier of vertical spindle mill, classification tests of coal were conducted in a lab-scale classifier, and its performance was evaluated by classification efficiency. Results show that yield and fineness of pulverized fuel (PF), as well as the classification efficiency of the whole size, would be improved by increasing the air volume. While the comprehensive classification efficiency reaches around 40% and keeps stable. Meanwhile, the separation phenomenon in the classification process was revealed by comparing the ash content between the feed and PF. Results show that ash content of each size fraction decreases after classification and presents growing trend with the decrease of particle size in PF. Separation effect is especially obvious for coarse particles, and its intensity increases as air volume reduces. Furthermore, such a phenomenon is also verified by industrial data, which proves again that minerals with high density would accumulate in the classifier reject. Deviations in bypass fraction, actual cut size, and comprehensive efficiency between regarding coal as homogeneous or heterogeneous were calculated. Thus, to be more accurate, density should be considered in the investigation of the coal particle behavior within classifier.