Abstract: Method and device for segmenting aligned characters. Each character is converted into a pattern of image elements. The image elements form columns and rows. The grey-values of the image elements are summed per column and the summed column values are put in relation with a fixed number of columns per character (1). After the outermost character position on a line has been found (2), a first segmenting process is carried out on the basis of a first probable distance between characters (3). After the segmenting process has been carried out for several different character distances (S.sub.1 -S.sub.n), the correlation factor is determined (6), by means of which the segmentation, which precedes the character recognition, is carried out.