We present a direct comparison between two different techniques for the relative calibration of time transfer between remote time scales when using the signals transmitted by the Global Positioning System (GPS). In the remote sites, the local measurements are driving either the computation of the hardware delays of the local GPS equipment with respect to a given reference GPS receiver, a “receiver” calibration, or the computation of a global hardware offset between two distribution reference points of the remote time scales, a “link” calibration. Both techniques do not require the same measurements on site, and we discuss the uncertainty budget computation differences. We report on one calibration campaign organized during Autumn 2013 between Observatoire de Paris (OP), Paris, France, Observatoire de la Côte d’Azur (OCA), Plateau de Calern, France, and NERC Space Geodesy Facility (SGF), Herstmonceux, United Kingdom. We show the different ways to compute uncertainty budgets, leading to improvement factors of 1.2 to 1.5 on the hardware delay uncertainties when comparing the relative link calibration to the relative receiver calibration.