It is vital to tobacco logistical company whether you can dig the potential high valued information from the substantial historical data that create during the tobacco products selling. This article adopts a more efficient algorithm that of fp-growth to design sales of tobacco products decision and support system, and the practice indicates the improved fp-growth algorithm is able to discover the sales discipline and relationship of tobacco products in various terms. In the algorithm, memory occupied is lower and respond speed is faster, which provides valuable references for tobacco logistical company on mastering sales discipline of tobacco products timely.