Task scheduling is a critical design issue for the service grid system. Decreasing the makespan of service and balancing the workload of the service resources are very important problems to be solved. Considering of the autonomous, heterogeneous, and distributed feature of the service grid system, we propose a novel task scheduling algorithm, called the adaptive Min-Min and Max-Min based on the non-dedicated constraint grid (NDA-MM) algorithm for the service grid. The proposed algorithm is provided with the dedicated property for the service resources and the paid property for the tasks, which is merged the high efficiency of the traditional Min-Min scheduling algorithm and the load balance of the traditional Max-Min scheduling algorithm. The experiments show that the algorithm overweighs traditional algorithms a lot in the makespan and the scalability of the scheduling algorithm.