Using a regional annular microwave array it was possible to produce a systemic temperature of 42 degrees C in approximately 80 min with applied net power levels of approximately 150 W. Resulting temperature distributions were non-uniform. Sites within the array were above systemic temperature during heating but approximated systemic temperature during the plateau phase. Sites outside of the array were lower than systemic temperature during heating and plateau phases. Dogs allowed to recover from the procedure experienced severe toxicity consisting of lumbar muscle haemorrhage, pain and swelling, and pelvic limb paralysis. Histologically, there was severe myopathy and haemorrhage and oedema in neural tissue in the caudal lumbar spine. Acute necrosis of lymphoid tissue was observed in all dogs. Temperatures in muscle reached 43-46 degrees C and were higher than at other measured sites. Spinal canal temperatures were essentially equal to rectal temperature, approximating 42-43 degrees C during heating and plateau phases. These data suggest regionally induced whole-body hyperthermia may result in: (1) power deposition non-uniformity leading to muscle and spinal canal temperatures which exceed systemic temperature and which are sufficient to cause serious toxicity; (2) systemic temperature non-uniformity which is undesirable for systemic thermochemotherapy; and (3) possible immunological dysfunction associated with lymphoid necrosis. Extreme caution must be exercised in administering energy to localized regions of human patients with the intent of elevating systemic temperature.