Long-term relationships are often governed by short-term contracts; this is usually explained by referring to the costs of specifying and enforcing a complete contingent contract. We focus here on the benefits usually associated with long-term commitment, namely the efficiency costs involved when long-term contracts are not available. We prove that renegotiable short-term contracts will implement the long-term optimum in a multi-period principal-agent framework when transfers are not too limited, objectives are conflicting, and there is no relevant asymmetric information at the contracting dates. This last assumption excludes adverse selection models, but not repeated moral hazard models when technologies and preferences are time separable.