With the rapid growth of China's economy and the steady progress of the urbanization, the speed of traffic facilities’ construction and improvement is far behind that of the increasing traffic demand. The traffic congestion problem has become increasingly prominent. A traffic signal control method based on Action-Dependent Heuristic Dynamic Programming (ADHDP) is investigated in this paper, and the algorithm of ADHDP is described in detail. The traffic signal controller based on ADHDP is designed. The algorithm is simulated on a two phase intersection with four entrance approaches, each of which has two lanes and each lane has a detector on it. Experimental results show that compared with traditional Fix-Time Control, the method has the ability of on-line learning, can effectively improve the average speed of vehicles and reduce traffic congestion.