The increasing scale and complexity of power grids exacerbate concerns about failure propagation. A single contingency, such as outage of a transmission line due to overload or weather-related damage, can cause cascading failures that manifest as blackouts. This paper articulates an approach for quantifying system survivability, which describes degraded operation in the presence of a fault in terms of 'system condition' and 'service capacity.' Also proposed is a recovery strategy based on importance analysis, where the goal is to maintain the highest survivability in the course of the recovery process. The proposed methods are illustrated by application to the IEEE 9-bus test system, a simple model system that allows for clear articulation of the process. Simulation is used to capture the effect of faults in both physical components of the power grid and the cyber infrastructure that differentiates it as a smart grid.