Overlapping activities are an effective means to accelerate the duration of Ｒ＆D projects． Project scheduling with overlapping modes and resource constraints is an extension of the classical problem． Firstly， based on an in-depth analysis of the influence of overlapping activities on project scheduling，a detailed de-scription and modeling of overlapping and its uncertainty is proposed． Then，a binomial probability model for reworking on downstream activities is built． Thirdly，an optimization model is built whose objective is to mini-mize the expectation of duration，and a serial schedule generation mechanism is designed as decoding operator for genetic algorithm． Finally，a heuristic method is used to test on the PSPLIB of J60 problems ( 480 cases) ． The duration of the project is analyzed as the network and resource parameters and overlapping parameters change． The results show that: the smaller the intensity of activities on resource requirements or the lower the scarcity of resources，the larger the number of overlapped activities，and the more the shortening in the project duration; Network complexity has little effect on shortening the duration; The more overlapping activities or the smaller reworking probability on downstream activities，the more obvious the impact of shortening the duration．