Climate shocks are argued to affect the likelihood of conflict through individual economic outcomes that change the individual opportunity costs of participation in violent activities. Studies testing this proposition, however, either fail to answer convincingly because they use aggregated data, because they rely on observed controls and strong assumptions about variable independence, or because their study sample is restricted to particular conflicts. This article uses two rounds in Afrobarometer where respondents were asked about participation in politically motivated violence as well as how their living conditions had changed in the last year. To get around endogeneity problems, perceived changes in living conditions are instrumented using a standardized precipitation-evapotranspiration index (SPEI). The study finds that participation in violence would, on average, have been more likely if an individual experienced a deterioration of living conditions than if it had not, in the subpopulation in Africa whose living conditions are affected by droughts.