Abstract: In view of the previous studies on prefabricated building scheduling with only a certain activity time and one execution mode for each activity, while actual scheduling processes are with uncertain activity time and various execution modes of activities, a fuzzy scheduling model with multi-objective and multi-mode resource constraints is established. An improved non-dominated sorting genetic algorithm (INSGA-II) is proposed to solve the (time-cost) bi-objective optimization model. The proposed algorithm performs population initialization and crossover operations according to the priority relationship of activities, meanwhile, a new three-segment coding method is developed containing activity, mode and resource lists. Finally, through the case analysis of an actual prefabricated building construction site and the comparison of algorithm performance, it is proved that the proposed scheduling model and algorithm can effectively solve the fuzzy scheduling problem under multi-mode resource constraints. It provides scientific ideas and methods for the design of construction scheduling plan.