There is disclosed a scanning electron microscope capable of removing the effects of vibrations on image information easily and reliably by detecting variations in the relative position between a specimen chamber holding a specimen therein and the specimen stage. The microscope has an image-processing portion that obtains information about the relative position between the specimen stage and the specimen chamber from a measurement unit when the beam is scanned. Based on the information about the relative position, a pixel position-correcting unit makes corrections to pixel positions indicated by the image information obtained by the scanning. An image creation unit creates image elements to eliminate pixel dropouts or pixel duplication produced by the aforementioned corrections. An image extraction unit extracts an image to be displayed.