Measurements made on the interstitial microwave antennas used for hyperthermia cancer therapy indicate that the heating patterns vary with the insertion depths (defined as the distance from the antenna tip to air-tissue interface). The antennas are made of thin coaxial cables with a radiation gap or gaps on the outer conductor. The antennas are inserted into small polypropylene catheters implanted in the tumour volume. This type of antenna may be simulated as an asymmetric dipole with one arm being the tip section consisting of the expanded extension of the inner conductor, and the other arm being the section of the outer conductor from the gap to the insertion point (air-tissue interface). We use four of the antennas to form a 2 cm x 2 cm array. The antennas are positioned on the corners of a 2 cm square. Measurements on both single antennas and multi-antenna arrays show that the maximum heating is not stationary with position along the antenna when the depth of insertion is changed. This paper investigates the theoretical prediction of the changes in heating patterns of interstitial microwave antennas at different insertion depths. Each of the antennas in the array is simulated as an asymmetric dipole. The SAR (specific absorption rate) is computed by using the insulated dipole theory. The temperature distribution in absence of perfusion is obtained through a thermal simulation routine to convert the SAR pattern into the temperature pattern. Excellent qualitative agreement is found between the theoretical heating pattern and the measured pattern in a non-perfused phantom on a 2 cm x 2 antenna array. Since the insertion depths of the interstitial antennas are different from patient to patient, it is recommended that simulation of the heating be done before treatments, to confirm the delivery of power to the target region.