For heterogeneous parallel and distributed systems with arbitrary processor release times, the make-span was minimized by finding the optimal distribution sequence of processors and the optimal load partition. First, the influence of processor release times on the make-span was analyzed with a given distribution sequence of processors, and the function of load partition with respect to the distribution sequence and time constraints was derived. Second, a new divisible-load scheduling model was proposed with the distribution sequence and time constraints as variables and the minimum make-span as the objective. Third, an effective global optimization genetic algorithm was designed to solve this model. Finally, experimental results show that the proposed algorithm outperforms the existing algorithms in finding the minimum make-span. © 2015, Editorial Board of Journal of Huazhong University of Science and Technology. All right reserved.