The ingredients for future geodetic reference frames can be separated into at least two categories, namely (1) theoretical, conventional, and definitional aspects, and (2) practical implementation aspects and physical components. In this Chapter, we focus on category (1). The present theoretical basis for the current ITRF is fairly simple at the moment (see, for example, McCarthy & Petit, 2004), which is appropriate as long as most of the infrastructure for the determination of the ITRF is ground-based. As more components of its realization and maintenance move away from the Earth’s surface into near-Earth space (and maybe beyond) and as accuracy “requirements” become more strict it is likely that closer attention will be paid to relativistic aspects. While the IAU has elaborated its recommended metric, some work with respect to ITRF determination and maintenance currently still uses Newtonian reductions with some general relativity “corrections” applied. This may no longer be an adequate approach in 2020. But also the Newtonian part of the theory is not fully at the targeted accuracy level. In particular, the theory of Earth rotation is especially weak with respect to increased accuracies. Moreover, motions of the RFO with respect to the CM are not explicitly accounted for, and higher degrees of surface deformation are neglected. Any coupling of the angular and linear momentum balance is not considered. As pointed out in Section 2.7, it is conceivable that by 2020 the international timescale will be maintained using ultra-stable frequency standards in orbit. Earthbased clocks, which are subject to large environmental “noise”, might only be used as local standards slaved to the orbital clock system. However, it is mentioned here that synchronization poses a severe problem which is not solved so far. Below, we do not consider relativistic effects. However, the concept and theory described below strives to provide a fully self-consistent system, which can then be the basis for a conventional system and frame. Currently, this is not the case in a strict sense and most likely also not at a perceptible level. There seems to be a