The temperature sensitivity of the proton-resonance frequency (PRF) has proven valuable for the monitoring of MR image-guided thermal coagulation therapy. However, there is significant inconsistency in reported values of the PRF-thermal coefficient, as measured from experiments encompassing a range of in vivo and ex vivo tissue types and experimental conditions. A method of calibrating the temperature dependence of the PRF is described and results are presented that indicate a tissue-type independence. To this end, other possible mechanisms for variations in the PRF-thermal coefficient are suggested, including physiological perturbations and volume magnetic susceptibility effects from geometry and orientation.