We re-examine the Miguel et al. (2004) study of the impact of growth on civil war, using growth in rainfall as an instrument. Miguel et al. (2004) - in our view, erroneously - include countries participating in civil wars in other states. Restricting the conflict data to states with conflict on their own territory reduces the estimated impact of economic growth on civil war. We show how spatial correlations in rainfall growth and participation in civil conflicts induce a stronger apparent relationship in the mis-classified data.