An image-processing instrument with an algorithm for the classification of etch pits such as flow pattern defects (FPDs), small pits (SPs) and large pits (LPs) revealed on silicon wafer surfaces after preferential etching was developed. In comparison with the usual human inspection, this instrument has the merit to obtain two-dimensional distributions of these pits in the wafer. Using this instrument, the pit distribution in two wafers taken from Czochralski (CZ) silicon grown at different growth rate was obtained. It was found that (i) the distributions of FPDs, SPs and LPs differed in each wafer, and (ii) the etch pit density changed with the growth rate of CZ silicon.