The invention relates to a disastrous consequence predication method of a CNG (compressed natural gas) filling station. The method comprises the steps of 1, removing background of an image by the differential method; 2, performing binarization processing for the image by the automatic threshold method; 3, performing median filter to remove various interference noises in the image; 4, calculating region area. According to the method, dangerous region correction factors are utilized, so that the result of the existing natural gas leakage disastrous consequence evaluation mode meets the actual condition; the influence of the barrier distribution and meteorological conditions is taken into account; FLUENT software is utilized to simulate the natural gas leakage dispersion condition, and the dangerous region correction factors are calculated according to the simulation result; the existing vapor cloud explosion quantitative evaluation model and API pub 581 quantitative consequence evaluation model are corrected through the correction factors, and therefore, the evaluation result meets the actual condition.