Abstract: An important topic of anti­money laundering is recognition of suspicious financial transaction.Directed primary circuit(DPC)is a very important structure in graph.Financial transaction data can be expressed by directed graphs，called financial transaction graphs，and it is obviously that DPC is suspicious if it exists in financial transaction graphs，because it means that money may return to starter.In this paper，we design a heuristic listing of DPC algorithm to help us to recognize these suspicious structures.The basic idea of the algorithm is to calculate strongly connected components(SCC)by using Tarjan’s strongly connected components algorithm at first，and then to conduct heuristic depth­first search(DFS)on every SCC which has been found.Differ to normal DFS，in the process of heuristic DFS，two heuristic information are utilized，one is parameter of direction to control the direction of DFS，and the other is in­degree and out­degree in SCC that are newly defined to choose visiting vertex by descending in­degree or out­degree.Also，we prove that there must be a DPC in a SCC which has 3 vertexes at least，and this proof provides theory basis to proposed algorithm.Certainly the time complexity of proposed algorithm is analyzed.Proposed algorithm reduces the size of query space，which leads to enhancement of algorithm performance.In order to test proposed algorithm，we choose several datasets of University of Florida Matrix Collection，and collect several practical datasets.All of datasets can be transformed to directed graphs.We design two experiments，one is for testing effectiveness of proposed algorithm，and the other is for validating necessity of two heuristic information.Experimental results show that our algorithm is effective，and utilization of two heuristic information is necessary.Thus，suspicious financial transactions，DPC，can be listed in financial transaction graphs，and technical assistance can be provided for anti­money laundering.