Developing a theoretically driven causal model of state failure is necessary if scholars are committed to improving the predictive power and forecasting ability of early warning models of state failure. Building upon the work of the State Failure Project and earlier statistical models of state failure, this study develops a causal model of state failure that is based upon a theoretical foundation that satisfies the parsimonious condition that policy forecasting models typically rely upon. The statistical model of state failure developed for this analysis extends the work of previous models by developing an alternative definition and measurement of state failure and introduces factors that are appropriate for making predictions and establishing causation.