Divergencies between chemical shift measurements of temperature and directly measured values using optical sensors have been studied in vivo in human peripheral muscle with the assistance of a variety of experimental and theoretical techniques. These include the modelling of both thermal and susceptibility changes using two- and three-dimensional finite element methods, as well as the use of multi-wavelength near-infrared observations. The conclusion of these studies is that a simple temperature calibration is not accessible, with results affected by the complex response of the tissue itself.