Character recognition systems, which employ a machine to recognize handwritten or printed characters, are prone to recognition errors in the cases of characters which are similar in shape. For example, the numerical character 7 is frequently mistaken for 9, and vice versa. In contrast, the characters 7 and 4 are not similar in shape, and 7 is unlikely to be mistaken for 4, or vice versa. Therefore, the rate of recognition error between these characters will be low. By considering this type of error model, this paper defines a new q-ary asymmetric error and proposes a method for constructing single q-ary asymmetric error locating codes. It also presents an evaluation of the code, demonstrating its superiority to the conventional single q-ary symmetric error correcting code. © 2003 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 87(3): 77–87, 2004; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.1156 [ABSTRACT FROM AUTHOR]