A tumour cloning system was used to cultivate breast cancer specimens. Fifty-six percent of 87 samples were adequate for evaluation, showing clonal growth in about one third (35%). Effusions yielded significantly better growth than solid specimens, the median colony numbers being 64 and 18 respectively. An attempt was made to examine whether there was any association between parameters accepted as prognostic factors for breast cancer and clonal growth in vitro. No correlation was found between preoperative tumour burden, histopathologic grading, menopausal status or overall survival and clonal growth in vitro, whereas we observed an inverse trend between progesterone receptor content of the tumours and their growth potential (P less than 0.01). In those few cases where in vitro and in vivo data could be compared, a high accuracy of the predicted sensitivities was found with respect to chemotherapy, but not in relation to hormonal treatment. A statistically significant higher overall chemosensitivity was associated with the absence of oestrogen receptors (P less than 0.01).