Computation grid originally aimed to combine supercomputers through internet so as to solve complex and large-scale computing problems. At present, the aim has evolved into combining various types of computers in different areas in a reasonable way and the highly integrated organic unity can supply powerful computing abilities to common users. Grid system consists of a wide variety of geographically distributed resources and these resources are heterogeneous, geographically distributed and dynamically available. High efficient scheduling algorithm would be able to increase throughput, maximize system utilization, and fulfill economical system and user constraints.This paper analyses existing resource scheduling strategy. A resource scheduling system model based on hierarchic with three-level for grid computing is designed, which consists of main scheduler, low-level scheduler and computing node. The main scheduler dispatches a portion of tasks to the low-level schedulers according to the requests of tasks and the execution performance of low-level schedulers, which actualizes the parallelism of task scheduling. In this paper, a hierarchic with three-level based feedback grid scheduling (HBFS) approach is presented. Experimental results demonstrate that this approach diminishes latency and contributes to the overall grid load balancing, which significantly improves resource utilization and response time of tasks.