This paper studies the application of multi-mode critical chain project scheduling problem based on genetic algorithm. A model of multi-mode critical chain project scheduling problem is addressed by combining the characteristics of multi-mode project scheduling problem and the critical chain method. Comprehensively considering the renewable resource constraints, nonrenewable resource constraints, activities with multi-mode and the uncertainty of activity duration, a bi-level genetic algorithm to solve multi-mode critical chain project scheduling problem is proposed. We use the shared benchmark from PSPLIB to test this new genetic algorithm and the results show that the algorithm in this paper has obvious superiority over most currently existing approach.