Along with the progress of communication technology and PC communications, the volume of mail order purchase is quickly on the increase, though actual direct purchase still takes up much weight. However, the systems so far have been data-based applications being run by PC communications such as HITEL and the like. They don``t much care for customers and do not give any suggestion and fail to put together the products of many different suppliers. We would develop a system as if a salesman were in the computer and sitting there to consult customers, suggest goods, and sell the product that they want. Observing the real world market, we classify the information for Sales Expert System to 5 categories: production information, customer information, constraints about customer and products, rules about customer and products, and salesman``s sales strategy. This situation goes with the framework of Constraint and Rule Satisfaction Problem (CRSP), so we develop a prototype system for men``s wear sales applying CRSP. During the application of it to our system, some features are added to those of CRSP in [Kwon 1993]. The result of the research will be used successfully for the future integrated marketing information system environment.