As the aggravation of environment pollution, increase of the number of private cars and the development of shared economic, one-way car sharing system is a substitute for traditional two-way car system. And free-floating car sharing system is most dynamic, in which users can rent and return vehicles in different stations just by smart phone. Therefore, no reservation information and imbalance between available cars and user demand of each station is a serious problem that lead to low user satisfaction and car working rate. In this paper, we produce a scheduling method generates short-term forecasting and relocation. In forecasting, we use Markov chain model （MCM） to forecast the number of available cars in future three time-period to catch the gap between it and demand. Then make relocate scheduling based on Genetic algorithm(GA) to minimize the gap. We apply this method to the data generated by simulator, randomly generating running condition, and we can find improvement in user satisfaction rate and car working rate.