Serum prolactin (HPr) has been measured in 459 patients 1 day before (HPr-1) and in 433 patients 10 days after (HPr-2) treatment. These came from an unselected sequence of 739 patients with operable breast cancer who had been referred to Guy's Hospital over a period of 5 yr. In addition HPr was measured in 100, or more, women at 3, 6 or 12 months after mastectomy. The median levels of either HPr-1 or HPr-2 were higher in pre-menopausal compared with postmenopausal patients (P = 0.03 and 0.06, respectively). Mastectomy was associated with increased serum HPr in both pre- and post-menopausal patients (P less than 0.001 in both cases). Average levels at 3 months, or after, were similar to those found before treatment. Nulliparous women had a higher median amount of HPr-1 than parous which was significant in premenopausal patients (P less than 0.008) whilst HPr-2 levels were not related to parity. Thus the rise in HPr associated with surgery was greater in parous than nulliparous women. Prolactin levels were not related to nodal status or tumour size. However, the amounts of HPR-2 were significantly greater in women with histological grade 3 tumours than those with grade 1 or 2. Standardising for either nodal status, tumour size or histological grade seven situations were found in which HPr-1 or HPr-2 levels were of prognostic significance. Although some of these significant associations could be fortuitous all shared a common feature that the least favourable prognosis was associated with the highest HPr levels.