A new statistical algorithm for the recognition of isolated hand written Arabic characters is presented. The algorithm is baaed on the use of a statistical technique known as the analysis of variance (ANOVA) as a feature extraction and classification technique. An (ANOVA)' classification structure known as the two-way classification was used to obtain the features of each character through the derivation of two F-statistics (F and F . ) rove column associated with the structure used. The two descriptors obtained, describe the distribution of character along rows and columns, respectively. A look-up table was set up containing the general descriptores for each character class and was used to classify an unknown pattern into one of the classes under investigation.