Abstract: To improve the Multi-Mode Resource-Constrained Project Scheduling Problem (MMRCPSP), based on constructed the mathematical model of MMRCPSP, a new genetic algorithm is proposed.The objective is to determine a mode and a start time for each activity such that all constraints are observed and the project duration is minimized.A special integer coding scheme is used in the algorithm, each gene represents the execution mode and the start time of an activity.By designing appropriate genetic operators, the proposed genetic algorithm is effective to solve this problem.The experimental result indicated the effectiveness of the algorithm.