The purpose of this study was to identify predictors of survival time in first recurrent breast cancer patients, including psychologic as well as biologic factors. Beginning in 1979, 36 women being treated at the National Institutes of Health for histologically proven recurrent disease were enrolled in this prospective study. At the time of data analysis, 24 had died from their malignancy. Through the use of a Cox proportional hazards model, four factors significantly entered the equation predicting survival time in the sample: Patients with a longer disease-free interval, who expressed more joy at baseline testing, who were predicted to live longer by their physicians, and who had fewer metastatic sites tended to live longer with recurrent disease than others in the sample (X2 = 22.9, p less than 0.0001). Findings from recent clinical and animal studies suggest that regulatory systems within the organism are linked and potentially influence one another. This study has demonstrated that factors at a number of levels--behavioral, as well as biologic--need to be considered in accounting for disease outcome variance.