Due to the uncertainty of future demand, companies have changed their production planning mode from pre-planning to produce according to the accepted orders, which is named as make-to-order (MTO) or build-to-order (BTO). The hypothesis of independent demand, which falls short of the real state, has been widely introduced in the model of traditional production planning. In general, the future demand and current orders are not completely independent. This paper introduces the data mining method of timing association rules to the model of MTO production planning. Through timing association rules, the relevancy of the future demand and current orders is considered and the future order demand can be forecasted from the history order data by association rules. The heuristic algorithm based on genetic algorithm (GA) is proposed to solve the model. A numeric example is built to demonstrate the proposed model and algorithm.