We have constructed an in silico system for the prediction of CYP2E1-mediated reaction using a two-dimensional template derived from substrate structures. Although CYP2E1 prefers small-size molecules for the substrates, the enzyme mediates oxidations of large-size molecules, such as benzo[a]pyrene. Overlays of these substrates, to assemble their sites of oxidation into a specific area, suggested a range of regions frequently occupied. The region, having a benzo[a]pyrene-like shape, was thus used as a CYP2E1 template. In this system, atoms in substrates, except for hydrogen atoms, were placed on corners of honeycomb structures of the template after having expanded the structures. Using published data for the metabolism on more than 80 substrates of CYP2E1, the core template was further refined to verify the adjacent area and to define the relative contribution of template positions for the catalysis. The positions on the template were classified into four different point (0-3) groups, depending on relative usage. In addition, we set independent points (-5 to 3) for specific positions to incorporate three-dimensional or functional information. Total scores from both position-occupancy and -function points were calculated for all the orientations of possible conformers of test substrates, and the scores were found to predict the relative abundance (i.e., order) as well as the regioselectivity of human CYP2E1 reactions with high fidelities.