A method for shortening a waiting time from the setting of a sample in a chamber to the stabilisation of the intensity for a secondary ion for SIMS analysis (mass analysis of the secondary ion) using a raster variation method is provided. By approximating so that the difference between time-lapse variations in intensities of the secondary ions sequentially measured for irradiation densities of two different primary ions becomes constant, a method capable of carrying out an accurate measurement of the concentration of an impurity in consideration of background noise despite time-lapse variations in the intensities of the secondary ions is provided.