‘The previous war in the Middle East was about oil, the next war will be about water’. Such predictions have been made regularly, and particularly with reference to the possibility of upstream-downstream conflicts in major rivers which cross interstate boundaries. A good case can be made that competition over water resources may exacer-bate conflict and contribute to interstate violence. More than 200 river systems are shared by two or more coun-tries and many of them lack enforceable water-sharing agreements. A number of rivers run between countries with a history of conflict, where water plays an important part in the economic life of the country. The dramatic state-ments about ‘water wars’, however, have a weaker foundation. As resource optimists have pointed out, there is an abundance of water where it is not subject to wasteful uses, human ingenuity can overcome water shortages, and nations can cooperate, rather than fight, to resolve international water issues. This study is built on newly generated data on boundary-crossing rivers, which have been added to the Correlates of War contiguity dataset. Our results indicate that a joint river does indeed increase the probability of militarized disputes and armed con-flict over and above mere contiguity. This risk factor is comparable in size to standard control variables, but much smaller than the effect of contiguity itself. Water scarcity is also associated with conflict, and the upstream-downstream relationship appears to be the form of shared river most frequently associated with conflict. But these results are not very strong and we do not have any systematic data on the issues involved in the shared-river conflicts.