Will climate change increase the prevalence of violent conflict in the future? To investigate this broad and difficult question, this thesis argues that we need to start with sound causal inference regarding the historical relationship between climate impacts on society and conflict outcomes. Earlier research reported highly disparate effects of climate shocks on conflict outcomes. This thesis makes a critical reading of the methodological approaches in these earlier studies, and suggests solutions to common issues endangering causal inference. A broad conclusion from the thesis is that the impacts of climate shocks on conflict are context dependent. Effects are only found for communal conflicts in resource-scarce areas where government services are lacking and where conflicts may be directly related to resources that are likely to be affected by changes in the climate. This insight becomes important when thinking about the relationship between climate change and the prevalence of violent conflict in the future. Supporting the positive socio-economic development that has occurred in the last 20 years in many poor and conflict-ridden states will be key to keep the impacts of climate shocks and climate change on violent conflict as low as they have been in the last decades.