This study investigates the use of Shapley values for evaluating the node significance in unreliable complex networks. We assume that in the network, node failures can trigger subsequent node failures, potentially leading to cascading failures. In such a network, the significance of a node cannot be appropriately assessed in isolation, as nodes are interdependent. Shapley values systematically provide an estimate of significance for each node by considering all possible combinations of node failures. An application of the Shapley value-based node significance is also presented.