This article analyses credibility and reputation in the context of peace negotiations. The model is applicable to the credibility of peace agreements in post-conflict situations in the immediate aftermath of civil war, where there is a danger of conflict breaking out again. The analysis is motivated by the empirical regularity with which peace agreements break down in the context of civil war, as for example in Angola, Sierra Leone and many of the conflicts in the Caucasus region. Where war provides economic gains to one side, peace is not incentive compatible, and peace agreements will necessarily degenerate, as they become time inconsistent. The party that has something to gain from surprise warfare, agreeing to peace but reneging on it, will do so and return to war. The levels of conflict chosen by this group are an increasing function of greed and capturable rents, but decreasing in the direct costs of war. In this context, action by external powers could lower the risk of civil war reigniting. Basically, these involve a greater commitment to peace, induced by policies known as commitment technologies. Sanctions, aid and direct intervention, if effective, could eliminate conflict, as well as help in devising commitment technologies to peace. In a multiple-time framework, uncertainty about the type of the group that may renege on peace generates extra costs in terms of greater fighting, and the key role of the discount rate in trading off present versus future consumption enters the analysis. High discount rates and the impatience to consume at present engender greater conflict.