Recognition of hand-printed characters has been intensively discussed during recent years. For office automation, forms should be automatically input to the computer instead of using humans to read and type the information. A character pattern is first scanned and digitized. A character written by hand, however, may have many possible shapes. In order to achieve efficiency and flexibility, a structural representation called hierarchical attributed graph representation (HAGR) is introduced to describe handwritten Chinese characters. The HAGR provides a simple and direct representation form for recognition of handwritten Chinese characters. With HAGR, the recognition process will become a simple graph matching. This approach can tolerate the variations of HAGR which reflect the instabilities or varieties of handwritten Chinese characters caused by different writing styles.