The purpose of this paper is to evaluate the safety and efficacy of deep hyperthermia in conjunction with radiation therapy. This study employed 'second generation' electromagnetic devices which were felt to be better able to confine heating and spare normal tissue than the devices evaluated in a previous study (RTOG 84-01). Sixty six patients at six institutions were enrolled on a prospective Phase I/II study. Eligible deep seated tumours were treated with a combination of external hyperthermia and radiation therapy. Radiation consisted of 1.7-2 Gy per fraction, 4-5 fractions per week, to > 20 Gy (previously irradiated lesions) or > 50 Gy (no previous radiation). Deep hyperthermia was delivered with electromagnetic devices: BSD 2000 for 92% of cases, Thermotron for 5% of cases, other low frequency electromagnetic for 4% of cases. Hyperthermia was delivered < or = twice weekly. Overall complete and partial response rates were 34% and 16% respectively. Response was not correlated with maximum tumour temperature or disease site. There was, however, a strong association with radiation dose: 54% CR with > or = 45 Gy versus 7% with < 45 Gy (p < 0.0001). The achieved temperatures were less than ideal. Although the average maximum tumor temperature was 41.9 degrees C (range 35.7 degrees C-46.7 degrees C), the minimum tumour temperatures were low. The average minimum tumour temperature was 38.5 degrees C and was never > 41.8 degrees C. Treatment was well tolerated with no fatalities. There were four acute grade 3 or 4 toxicities (6% of patients). Patient discomfort resulted in interruption or discontinuation of sessions in 30% of the sessions. In 12 cases (18% of patients) the planned course of hyperthermia was discontinued due to acute discomfort. The devices used in this study were better tolerated than the devices used in the previous Phase I/II deep hyperthermia trial (RTOG 84-01) with less patient discomfort and no problems with severe systemic cardiovascular stress. In the previous study 68% of the hyperthermia courses were prematurely terminated primarily due to patient discomfort and toxicity; in the present study 18% were prematurely terminated. However, as indicated by the low minimum tumour temperature, fundamental problems with achieving acceptable temperature distributions remain.