A Phase I/II study was undertaken to investigate the efficacy and side effects of combined hyperthermia and radiation therapy in the management of presumed or known microscopic residual tumors. Between February 1985 and March 1991, 262 fields in 89 patients with local-regional recurrent breast cancer were treated with externally administered hyperthermia and radiation therapy. Thirty-eight fields were treated for microscopic residual disease following excisional biopsy of nodular recurrences and 224 fields were treated electively for areas at high risk for local recurrences adjacent to fields with macroscopic residual disease. Mechanically mapped temperatures were monitored throughout the field in all treatments. All patients had at least one follow-up evaluation at three weeks or more following completion of treatment. The majority of the fields were in patients who had had extensive prior therapy including radiation therapy (54%), chemotherapy (71%), and hormonal therapy (51%). All fields received hyperthermia (1-6 treatments: average 1.74) and radiation therapy (average dose: 42.4 Gy); concurrent hormonal therapy was administered in 37% of the treatments and no fields received concurrent chemotherapy. The treatments were well tolerated, no life-threatening complications were noted. Averages for all fields of the minimum, maximum, and average measured interstitial temperatures were 40.2 degrees C, 45.3 degrees C, and 42.8 degrees C, respectively. The three-year actuarial local-control rate for all 262 treated fields was 68%. Parameters characterizing the initial breast cancer, the patient and tumor at the time of hyperthermia, and the treatment were studied in univariate and multivariate analysis for correlation with duration of local control within the hyperthermia treatment field. Parameters in the best five covariate model correlating with the duration of local control included: estrogen receptor status of the initial breast cancer; initial T-stage; time from initial breast cancer to first failure; age at hyperthermia; and concurrent radiation dose (p-value for model less than 0.000001). Six covariate models adding anatomic site of disease, field type, mean minimum temperatures, and mean percent temperatures greater than or equal to 40 degrees C all resulted in improved models. Randomized controlled studies stratifying for these pretreatment parameters are felt warranted to confirm the value of adjuvant hyperthermia in the elective treatment of areas of high risk for local-regional recurrent breast cancer and in fields following surgical excision of recurrent disease, particularly in patients in whom full dose radiation therapy cannot be safely administered.