The availability and reliability of large critical infrastructures depend on decisions made by hundreds or thousands of interdependent entities and, by extension, on the information that the entities exchange with each other. In the electric power grid, the widespread deployment of devices that measure and report readings many times per second offers new opportunities for automated control, which is accompanied by the need to automatically assess the trustworthiness of the received information. This paper proposes a Bayesian estimation model for calculating the trustworthiness of entities in the electric power grid and making trust-relevant decisions. The model quantifies uncertainties and also helps minimize the risk in decision making.