A hybrid genetic scheduling strategy (H-GA) is described in this article, H-GA combines with grouping and load balancing strategy based on traditional genetic algorithm (GA). First, tasks are divided into several different subgroups by task granularity. Then, task subgroup which is selected by granularity from big to small is used to schedule by the genetic algorithm, and during scheduling, the load balancing strategy is used to adjust task distribution in the individual. Grouping can cut down the length of individual, which speeds up convergence of genetic algorithm. Load balancing strategy can make the individual better, which also speeds up convergence of genetic algorithm. The implementation shows that converging speed of H-GA is faster than GA, and result of H-GA is optimal than GA if the iteration times are equal.