To adequately plan and administer localized hyperthermia with interstitial microwave antennas, the thermal distribution patterns generated by such antennas must be characterized. This study evaluated the performance of single node 915 MHz antennas operating either alone or as a 2 cm square array of four parallel antennas using both muscle-equivalent phantoms and canine thigh muscle. Two types of measurements were compared. Specific absorption rate (SAR), where temperature increases resulting from short duration microwave pulses were used to define power distribution, and temperature gradients during simulated hyperthermia treatments. SAR measurements in phantoms were comparable to those obtained in non-perfused canine muscle demonstrating the usefulness of the phantom for these measurements. For a single antenna there was a rapid decrease in power radially which resulted in a steep thermal gradient at distances within 0.5 cm. However, the power generated by a four-antenna array was highest along its central axis and declined to approximately 50% near the antennas at the array periphery. Along the central axis of the array power decreased most rapidly distal to the antenna nodes. The distribution of temperature measured during simulated hyperthermia treatments in phantoms paralleled the SAR distribution and was comparable to the temperature gradient observed in perfused canine muscle, suggesting that phantoms could be used to predict temperature distributions in resting muscle tissue.