This prospective trial was designed to help in selecting therapy for patients with elevated and normal plasma prolactin. Ninety-two patients entered this trial, of whom 86 were evaluable for final analysis. Hyperprolactinemic patients (n = 31) were randomized to receive VAC/FMC chemotherapy with or without bromoergocryptine. Normoprolactinemic patients with ‘low risk’ metastatic disease (disease-free interval > 30 months, ER/PR positive or unknown) were treated with medroxyprogesterone acetate or VAC/FMC chemotherapy. Normoprolactinemic ‘high risk’ patients (n = 42) (disease-free interval < 30 months, ER/PR negative) received VAC/ FMC chemotherapy with or without medroxyprogesterone acetate (MAP). The results show that bromoergocryptine does not improve response rate, duration of response and survival. Median survival of patients with elevated basal plasma prolactin ( > 15 ng/ml) is reduced to 9 months compared to patients with normal basal plasma prolactin (17 months, log-rank p = 0.005). Unexpectedly, TRH stimulation proved inappropriate to separate normo- and hyperprolactinemic patients in terms of survival. Normoprolactinemic ‘low risks’ (tamoxi-fen failures) were observed to qualify for further hormone therapy (median survival 21+ months). Normoprolactinemic ‘high risks’ showed median survival of about 12 months with no apparent benefit . for those receiving MAP, additionally. The results suggest that basal hyperprolactinemia, disease free interval, ER/PR receptor status, and liver metastasis are important prognostic variables. Endocrine and cytotoxic chemotherapy should be selected according to these risk factors.