According to the current imperfections in the study of low-permeability reservoir classification，a low-permeability block in Xinjiang was selected as an example. Firstly，the classification parameters were selected. Secondly，combined with the distribution of the actual reservoir classification parameters，preliminary reservoir classification boundary of this block was given. Finally，the reservoir classification parameters were characterized and analyzed comprehensively by the reservoir numerical simulation technique，and the influence degree of each parameter on the productivity was evaluated，which are in the descending order as follows：threshold pressure gradient，permeability，pressure sensitivity factor，effective porosity，oil viscosity，movable fluid percentage，mainstream throat radius and clay mineral content in ascending. By assigning weights to each parameter according to the influence intensity，the reservoir comprehensive classification coefficient was constructed，an eight-element comprehensive classification method was proposed. This method was applied to draw the scatter diagram of actual single-well production distribution in the target block. It is found that the reservoir comprehensive classification coefficient is consistent with the single well production of the block，and the accuracy of the method has been proved. Finally，according to the on-site classification criteria for the target block based on oil production capacity，the reservoirs were divided into 4 types based on the level of productivity using this method.