There have been controversy and uncertainties surrounding the rotational behavior of conservative internal bending moments in three-dimensional frames. The two contesting representations of conservative internal bending moments are the quasi-tangential and the semitangential moments, each of which has its own shortcoming from the purely physical point of view. This situation must be rectified, as the correct identification of the nature of internal moments is indispensable for reliable buckling and large displacement analysis of three-dimensional structures. This paper presents a new type of conservative moment that has hitherto never been considered in the literature, and demonstrates that the proposed representation of conservative internal moments can be justified on both physical and rigorous mathematical grounds. Various issues relating to finite rotations and the work done by conservative moments in space are also addressed.