Short-term traffic flow forecast is an important issue in Intelligent Transportation system. Due to the low forecast accuracy of exponential smoothing method in data fitting. In this paper, By combining exponential smoothing theory and Markov chain, we propose exponential-smoothing-Markov short-term traffic flow forecast method. With Markov chain, the method can solve the problems existing in exponential smoothing, i.e., by narrowing the forecast interval and improving status weighting centers and status transition probability matrices in pre-calculation, the proposed method can improve the future status forecast accuracy. An experiment on actual traffic flow has shown that, the proposed method improves forecast accuracy and has stronger adaptability.