Neomalthusians have regularly predicted "water wars" while cornucopians have argued that there is no inherent water scarcity and liberal institutionalists have seen cooperation as a more likely outcome of competition for limited water resources than violent conflict.
Three earlier studies have found a positive statistical relationship between shared rivers and low-level interstate conflict. Based on a more comprehensive dataset, an improved model of conflict, and a more appropriate control for geographical opportunity, we argue that these results are spurious and that we cannot establish a conflict-inducing effect of shared rivers over and beyond contiguity itself. In fact, the new dataset presented here makes it clear that nearly all neighbors in the international system share at least one river. This calls for a different approach to investigating interaction in shared river basins. This notwithstanding, freshwater and other shared resources may still provide a mechanism to explain why contiguity is robustly associated with conflict, so the water-conflict scenario cannot be dismissed. Indeed, our results show that among river-sharing states, basins with an upstream/downstream configuration increase the risk of conflict. The article finally discusses how river interaction should be further investigated based on these results and what new data are needed to enable such research.