Pattern recognition is an essential part of any high-level image analysis systems. Arabic language has four forms for each letter depending on the position of the letter in each word. These are initial, medial, final and isolated. This research concerned with recognizing an isolated Arabic Characters using Hu's seven moments concept. Moments and functions of moments have been extensively employed as invariant global features of images in characters recognition. Character recognition performed by two stages; first is the training stage and the second is testing stage, these involves several major functions starting from the input character until deciding the recognition of character. These functions are preprocessing, feature extraction and characters matching. Preprocessing function includes; image acquisition, noise removal using median filter, image binarization and characters segmentation. The characters matching include the computation of the Euclidean distance between testing and training characters. Recognition system has been performed on the printed Arabic characters and the percentage accuracy for recognize 65 printed characters was 96.92307% While in hand written characters the recognition percentage is decreased, due to irregularities appears in the handwritten characters.