The assessment of a non-invasive technique for measuring deep body temperature is described. An experimental system, in which the insulating effect of subcutaneous fat thickness could be simulated, was designed to test the deep body thermometer (DBT). The assumption that the instrument produced zero temperature gradient was found to be incorrect, and a mathematical model was developed which showed that the cyclically varying output of the device is an inherent property of its design. The accuracy and response time were investigated as a function of tissue thickness to assess the instrument's clinical usefulness.