Artificial immune system (AIS) based classification approach is relatively new in the field of pattern recognition (PR). This paper explores this paradigm in the context of a frequently occurring PR problem, namely discrimination of similar shaped character pairs. The problem has been studied in the context of improving the recognition accuracy of OCR (Optical Character Recognition) systems that often make mistakes to properly classify the confusion pairs. A set of binary classifiers is designed following immune principles to achieve pair-wise discrimination. The performance of the proposed approach has been investigated in detail and compared with classification schemes like nearest neighbor and support vector machines (SVM)-based approach.