Dimethylbenzanthracene-induced rat mammary tumors were defined as either prolactin responsive or prolactin independent on the basis of growth response to prolactin administration. There was no difference in tumor binding of prolactin between the two groups when tumors were biopsied before treatment. Prolactin binding was, however, significantly higher in responding tumors when biopsies were obtained following treatment. By contrast, when tumors were defined as responsive or independent on the basis of response to suppression of serum prolactin with bromoergocryptine, there was significantly higher prolactin binding in the responsive than in the independent group both before and after treatment. During serial treatment with prolactin followed by bromoergocryptine, there was a progressive decline in prolactin binding to tumor biopsies, particularly in prolactin-independent tumors. Prolactin binding to pretreatment tumor biopsies thus did not predict which tumors would respond to administration of prolactin but, for the total group, did indicate tumors likely to regress with prolactin withdrawal. However, the correlation between prolactin binding and tumor regression following hormone withdrawal was not sufficiently strong to permit reliable prediction of behavior for individual tumors. Prolactin-independent growth was associated with decreased prolactin binding to tumor tissue, particularly following manipulation of serum prolactin levels.