Models have been developed for the signal changes occurring in human peripheral muscle after thermal stress for some methods of measuring temperature in vivo. These include the dependency of T1 on temperature, observed by both direct, and magnetization transfer, experiments, and the temperature sensitivity of the diffusion coefficient and the chemical shift of the water proton line. Results show the relative complexity of the T1-dependent models when there are significant variations in, possibly amongst other things, tissue perfusion, and how the predictions made using them match the behavior of the practical values quite well. Using changes in the diffusion coefficient appears less problematic, although there can still be significant errors in temperature calibrations. The problems with this approach are likely to be due to the consequences of anisotropy in tissue.