The resources in the multi-skilled project scheduling problem(MSPSP) are assumed to be multi-skilled. Compared with the single-skilled project scheduling problem, the resource allocation has more influence on the makespan in the MSPSP. Therefore, more attention should be paid to resource allocation when scheduling a multi-skilled project. In this paper, we propose a heuristic algorithm to solve the MSPSP from the perspective of resource allocation. The algorithm uses a maximum matching method of bipartite graphs to select schedulable activities and develops a dynamic resource priority to allocate resources based on the parallel scheduling scheme. The core idea of the algorithm is to retain resources which are more flexible or more important to the subsequent activities for the next decision time. In order to verify the effectiveness of the proposed heuristics, a series of computational tests are conducted. The results show that the proposed algorithm has a significant advantage over the algorithms based on random or static resource priority.