There are a number of unreliable resources in grid market, which correspondingly bring to resource service risk. It obviously affects grid scheduling and utility seriously. In this paper, a classified approach is proposed to divide resource service risk into tangible risk and intangible risk according to how the resource service risk engenders. Then it introduces the models of tangible risk and intangible risk respectively and presents a normalized approach to evaluate resource service risk. A grid resource service risk based scheduling algorithm framework is proposed at last. Compared with traditional scheduling algorithms, the service risk based scheduling algorithm efficiently bestirs the reliable resource suppliers and reduces the job failure rate and total costs obviously in grid market.