As the demand for seasonal products change frequently, it is crucial to form an accurate estimation of seasonal demand prior to its actual sales. Accurate demand estimation can lead to higher customer satisfaction and lower operational costs. The purpose of this dissertation is to improve the quality of a market sales forecast in order to improve the effectiveness and efficiency in a company. This research examines the monthly market sales forecast of a seasonal demand product in Taiwan, the split type air conditioner. Most market sales have a pattern including both seasonality and trend. Consequently, a comparative study is conducted on Winters multiplicative trend seasonal model, Winters additive trend seasonal model, the Decomposition model, the equal-weight combining method, and the expert systems. The results show that the expert systems rule 1 has the smallest mean square error, while expert systems rule 2 has the largest mean square error for the period from January, 2010 to December, 2010. In summary, we conclude that the expert systems rule 1 is the simplest and the most accurate method in this case.