In the Service Oriented Grid, a grid application can be composed of diverse Grid Services which may be already deployed on grid sites, and dependencies between these Grid Services can be described as the workflow graph. Though many grid scheduling methods have been developed, they have been studied based on static heterogeneous computing environment. For this reason, these heuristics cannot be directly applied to Grid, which is more dynamic computing environment. Therefore, we develop the dynamic service scheduling method, which support that the workflow applications can be performed effectively in spite of dynamic Grid nature. Lastly, we show that our proposed scheduling method outperforms than other conventional scheduling methods through the various experiments.