A Phase I-II trial testing the addition of systemic cisplatin (CDDP) to local hyperthermia and radiation was conducted to determine the dose of cisplatin that is tolerable once weekly for 6 weeks and to estimate the therapeutic potential of this trimodality combination in patients with locally advanced malignancies. Cisplatin at 20 mg/m2 (4 patients), 30 mg/m2 (8 patients), and 40 mg/m2 (12 patients) was given rapidly (over 5-10 min) i.v. after prehydration with 1 liter of normal saline. After approximately two-thirds of the cisplatin dose had been delivered, microwave hyperthermia was begun and continued for 60 min; the target minimum tumor temperature was 43 degrees C. Following hyperthermia, a 400 cGy fraction of radiation was delivered to the tumor. On other days during the treatment weeks, additional 200 cGy fractions were given to total doses of 6,000-6600 cGy in patients with full radiation tolerance or 2400-3600 cGy in patients with limited radiation tolerance. The 24 patients in this trial had a median age of 57 years and the predominant sites/tumor types were head and neck/squamous cell carcinoma (9) and chest wall/breast adenocarcinoma (9). Seventeen of the 24 treated tumors (70%) had previously been irradiated. Eighteen patients (75%) had received prior chemotherapy and nine patients (38%) had previously been treated with cisplatin. Bone marrow suppression was dose limiting in patients heavily pretreated with chemotherapy and chest wall radiation. No significant toxicities were observed at the 20 and 30 mg/m2 dose levels, but 5 of the 12 patients (42%) treated at 40 mg/m2 required modification of the cisplatin dose because of blood count suppression in four patients and mild renal dysfunction in one patient. Each of the patients with bone marrow suppression, however, had been heavily pretreated except for one patient with thrombocytopenia due to hypersplenism. Nausea and vomiting were mild with use of a standard, multiagent antiemetic regimen. Twelve patients (50%) attained a complete regression (CR) and 12 patients (50%) a partial regression (PR). Complete regression appeared to correlate with small tumor volumes (115 cc for CR versus 199 cc for PR patients) and higher tumor temperatures (4.6 average minimum equivalent minutes at 43 degrees C in CR versus 2.0 min in PR patients). Local toxicities included second degree burns in 12 patients (50%) and third degree burns in 6 (25%), but all burns healed in 4-12 weeks without surgical intervention.(ABSTRACT TRUNCATED AT 400 WORDS)