To provide a method for quantitatively evaluating surface defect which can quantitatively evaluate the surface defect of a molded resin product and, in addition, can make the evaluated value to agree with that of an inspector.

A stripe picture on the surface of a bumper is taken at a low angle with a sensor head 100 and fetched to a personal computer 300 and a binary picture is acquired by binarizing the fetched stripe picture (original picture). Then two characteristic amounts (the peak value of a power spectrum and the overall value of the results of FFT (fast Fourier Transform) are extracted by performing FFT on all lines of the binary picture in the lateral direction and, at the same time, another characteristic amount (the ratio of the standard deviation to the mean value) is extracted by finding the stripe widths on all lines from the picture obtained by filtrating the binary picture. Then the flapping rank (the evaluated value of the surface defect) of the surface of the bumper is discriminated based on the acquired three characteristic amounts.