The stray radiation produced by microwave applicators during clinical diathermy treatments creates a potential health hazard both to the patient and to the therapist. It is important to quantify levels of stray radiation so risk of excessive exposure can be minimized. In this study, tissue-substitute models—a planelayered block, a large cylinder, a thigh-model cylinder with one flat side, and a small cylinder—were irradiated by a 915-MHz direct-contact applicator, a 2450-MHz Transco circularly polarized direct-contact applicator with choke, and a 2450-MHz “C” director. Power-density measurements were made in proximity to these models to determine levels of stray radiation. Human volunteers were also irradiated by the 915-MHz direct-contact applicator and by the 2450-MHz Transco applicator. The human anatomical sites irradiated were chosen for their geometric similarity to the models. Measurements made near the upper and lower areas of the dorsum were compared with those of the plane-layered model; measurements near the thigh were compared with those near the large cylinder and the thigh model; and measurements near the biceps were compared with those made near the small cylinder. When the direct-contact applicators were not in complete contact, levels of stray radiation above 5 mW cm−2 were measured at points 5 cm from the external surface of the applicator. The noncontacting “C” director had stray-radiation levels above 5 mW cm−2 at 5 cm from the applicator in all directions. When the E field was perpendicular to the long axis of the muscle or the model, higher levels of stray radiation were measured. It was found that the models used were of limited value in predicting patterns of stray radiation near the human counterparts.