Classical Bellman-Ford algorithm is improved to solve the shortest path problem with bounded edge number efficiently. Using the experience of partitioning algorithm for reference, two improved algorithms are obtained, which can decrease the number of distance labels of vertices. Since all existing improved Bellman-Ford algorithms can’t solve the shortest path problem with bounded edge number, these two improved algorithms are entirely new. In contrast to the common version of Bellman-Ford algorithm, these two improved algorithms can save storage space efficiently, and can raise computing efficiency remarkably.