This paper describes a system being developed to recognize date information handwritten on Canadian bank cheques. A segmentation based strategy is adopted in this system. In order to achieve high performances in terms of efficiency and reliability, a knowledge-based module is proposed for the date segmentation and a cursive month word recognition module is implemented based on a combination of classifiers. The interaction between the segmentation and recognition stages is properly established by using multi-hypotheses generation and evaluation modules. As a result, promising performance is obtained on a test set from a real-life standard cheque database.