In this paper we study the response of vote shares to economic fluctuations and conflict. Spain seems to be the ideal niche for a case study like this since it has experienced both phenomena during the last decades. Recent Spanish democratic history has witnessed four complete economic cycles, with deep recessions and pronounced booms. During this period, there has been a nationalistic conflict with terrorist manifestation. We use Spanish provincial data from the ten congressional elections since the end of Franco's dictatorship. Vote shares at provincial level are modeled as fractional responses to unemployment, inflation, terrorism assassinations, turnout and other factors. The statistical model used, a fractional probit, specifies conditional means of district and election unobserved effects as linear functions of the covariates. Estimates of National Partial Effects (NPE), i.e. the effect on national vote shares of changes in unemployment, inflation and terrorism are statistically significant and quantitatively important. In addition, vote shares respond to participation rates and these also depend on economic factors and terrorism, thus creating an endogeneity problem. The expected margin of victory is then used as instrument for turnout.