Within the framework of autonomous morality, the validity of a norm depends on the consent of autonomous agents. However, such a consent will often not be obtained. Approximative criteria can be formulated, stating when a norm, although not consented completely, should be accepted as objectively valid. Such criteria, however, restrict moral autonomy. Hence, demands of autonomy and of rationality collide. This collision must be resolved by restraining autonomous judgement in order to make commonly binding norms possible.