This report examines the bases for developing radiofrequency exposure standards which can be related to the thermogenic properties of electromagnetic fields. A review of selected biological effects, including dosimetric data and simulation of human thermodyanmic characteristics that are pertinent to standards development, is presented. Based on the analogy of thermal-stress standards that have been developed for hot industrial environments, limits on increases of body temperature are proposed as criteria for limiting exposure to radiofrequency fields, i.e., occupational exposures involving deep heating of the whole body should not increase core temperature in excess of 1 degree C. Since energy deposition from exposure to some RF fields is likely to be non-uniform and may be high in tissues that are not adapted to high rates of absorption or dissipation of thermalizing energy, means are needed to adjust focal thermal loading against the whole-body averages. A limit on core temperature is inadequate when focal elevations of temperature are close to the limits for protein denaturation, as may well occur even though the core temperature may rise less than 1 degree C. Safety limits for the general population are also discussed and here the permissible thermal load should be low enough to cause no more than an insignificant increase in core temperature. Areas needing further research to reduce the uncertainties in developing safe exposure limits for man are delineated. Even in highly adverse environmental conditions the gross thermal load and consequential heat stress from exposure to radiofrequency fields at the 10 mW/cm2 level will be small compared with that generated by any physical effort. On the basis of available data, it is concluded that the safe value for continuous exposure to 10 mW/cm2, widely used in Western countries, appears to provide an adequate margin of safety for both occupational and environmental exposure for frequencies above about 1 GHz. This limit may well be too high (perhaps by an order of magnitude) for some frequencies below 1 GHz where body resonances cause a significant increase in energy deposition and where local temperature rises occur. At the same time the present averaging period of 0.1 h seems unjustifiably short.