Hyperthermia greater than or equal to 42 C is tumoricidal, apparently a function of absolute temperature and duration of heating. However, because the technology for producing safe and effective deep hyperthermia did not exist, there was virtually no data on the thermal death times of human cancers. The development of a fundamentally new radio frequency device that produces uniform hyperthermia to any depth without surface tissue injury has allowed preliminary testing of this hypothesis in 38 patients with advanced cancer. Temperature measurements were taken in tumors and normal adjacent tissues. Tumors were heated from 40 C to greater than or equal to 50 C, one to ten times (13-600 minutes). Serial tumor biopsies compared percent necrosis (total absence of nuclei) by type of therapy. Of 44 tumors evaluated (21 superficial, 23 visceral), 31 (70%) were heated greater than or equal to 42 C, 23 (52%) greater than or equal to 45 C, and 14 (32%) greater than or equal to 50 C, with virtually no normal tissue injury. Single, short duration hyperthermia at greater than or equal to 50 C resulted in 20-100% tumor necrosis, while lower temperatures had no effect. Two or three treatments at 45-50 C produced 70-100% necrosis, while lower temperatures produced less necrosis at more than twice the duration of heating. Multiple treatments produced increased necrosis at lower temperatures; however, at the same temperature duration, higher temperatures were most effective. These clinical results support the hypothesis that the observed necrosis is related to both temperature and treatment time and suggest that higher temperatures and longer durations of therapy are most beneficial.