Spinning reserve (SR) is one of the most prevalent methods for balancing grid uncertainties, such as generator faults, to maintain grid reliability. Literature review shows that several deterministic as well as probabilistic methos have been proposed for determining SR. It is always a challenge for a system operator to decide which approach better from security and reliability point of view. This is important because the allocated SR may provide in some cases a misleading sense of confidence with respect to safe, secure, reliable and economic operation of power systems. This paper presents a cost-based risk index approach for assessing the spinning reserve requirements in a power system. To that end, the performance of spinning reserve is classified in three types, namely, not-effective, partially-effective, and not-meeting-load. Then probability of each type and its consequences are subsequently computed and finally that the risk associated with any spinning reserve value is determined. It is shown that one might consider various spinning reserve values for an operating condition (randomly or using approaches proposed in literature), then calculate risks associated with each value, and finally use the calculated risk indices to determining the optimal level of spinning reserve. As an example, we have shown that in the studied network with 6600MW load, maintaining 240MW SR will increase cost of 1MW hour energy by $0.154 while the optimal value of 200MW SR will increase cost of 1MWhour energy by just $0.126. This paper initially focuses on providing a measure of the quality of an ex-ante specified spinning reserve, latter on the flowchart of using the proposed approach for determining optimal level of spinning reserve is presented. The proposed risk index can also be used for comparing different deterministic as well as probabilistic approaches presented in literature for spinning reserve requirements.