The vast majority of existing studies on data distribution in vehicular networks utilize the mobile vehicular nodes as content carriers, however the high mobility, the limited resources and the security risks of vehicular nodes prohibit the performance improvements. This paper proposes a distributed storage scheme named DSS using bipartite graph matching for vehicular networks where roadside units are regarded as storage devices. Considering the content replication at roadside units is NP-complete, DSS tackles the problem by transforming it into a maximal matching problem of bipartite graph in which the left vertices stand for the requests from vehicular nodes and the right vertices represent the storage cells of roadside units. Hence the original problem is solved by Hungarian algorithm in polynomial time. In addition, a redundant content deletion algorithm overcomes the possible duplications generated from the problem transformation by checking the redundancy in the order of the response factors of different replicas. Experiments show that the DSS scheme achieves a high access ratio and keeps a short access delay with a small cost of network resources.