How substantial is democracy as a cause of transnational terrorist attacks? Can our identification of democratic political systems help us to anticipate the flow of transnational terrorism? We seek to address these questions by analyzing data on transnational terrorist incidents from 1968 to 2007. We rely on receiver operating curves as a diagnostic tool to assess forecasting ability of various models of terrorist activity. Our analyses yield four central conclusions. First, our model of transnational terrorism provides a fairly strong basis for forecasting attacks—at least at the (relatively broad) level of the country-year. Second, while the overall forecasting capacity of this model is fairly strong, democracy adds very little to our capacity to forecast terrorist attacks relative to a parsimonious model that includes only distance and the prior history of terrorism. Collectively, these two variables perform about as well as a much more broadly specified model in forecasting terrorist attacks out of sample. Third, the model is highly redundant in a predictive sense. That is, many if not most of the other variables appear to provide similar information in terms of identifying terrorist attacks. Finally, we suggest that scholars focus on the development of more fine-grained and time-variant predictive indicators in order to improve our ability to forecast transnational terrorism.