The problem of measuring a spatially varying specific absorption rate (SAR) by thermal techniques is treated both analytically and numerically. It is shown that, unless the measurement is attempted at an inflection point of the SAR distribution, it will be confounded by thermal diffusion within the medium. Rules of thumb are provided to enable the experimenter to gauge the thermal conduction contributions (i.e. error) to the thermally determined SAR near a spatial extremum. The simplest of these is that the width t(m) [s] of the time window, over which temperature variation associated with SAR is measured, should satisfy the inequality t(m) </approximately alpha(2)/pi(2)D, where alpha [m] is a characteristic length scale of the spatial variation of the SAR and D [m(2)/s] is the thermal diffusivity of the medium. The above considerations are valid, independent of the source of the deposited energy.