Cooperation occurs even where it is not predicted by economic theory, owing to what is widely recognized as too narrow a conception of self-interest. In particular, relying on plenty of experimental evidence, it has been maintained that agents adopt such a strong reciprocity rules in their behavior as make it worthwhile to punish those who defect or do not act fairly, costly though as this may be. We propose to lay the analytical foundation of such behavior – and more generally to cooperation-proneness – by considering selfesteem. Agents may include self-esteem in their utility (or goal) function and actually produce or destroy self-esteem through their behavior. This amounts to introducing a moral system in individual behavior in such a way as to make it amenable to rational maximization. We also show how the impact of self-esteem on the best contract in Principal-Agents situations and how such impact differs in Moral Hazard and Adverse Selection situations.