The differentiation of a pituitary non-functioning macroadenoma from a macroprolactinoma is important for planning appropriate therapy. Serum PRL levels have been suggested as a useful diagnostic indicator. However, values between 2500 and 8000 mU/l are a grey area and are currently associated with diagnostic uncertainty. We wished therefore, to investigate the serum PRL values in a large series of patients presenting with apparently non-functioning pituitary macroadenomas. All patients presenting to the Department of Endocrinology in Oxford with clinically non-functioning pituitary macroadenomas (later histologically verified) between 1990 and 2005 were studied. Information documented in the notes on the medications and on the presence of conditions capable of affecting the serum PRL levels at the time of blood sampling was also collected. Two hundred and twenty-six patients were identified (median age at diagnosis 55 years, range 18-88 years; 146 males/80 females; 143 gonadotroph, 46 null cell, 25 plurihormonal and 12 silent ACTH adenomas). All tumours had suprasellar extension. At the time of blood sampling 41 subjects were taking medications capable of increasing serum PRL. Hyperprolactinaemia was found in 38.5% (87/226) of the patients. The median serum PRL values in the total group were 386 mU/l (range 16-3257) (males: median 299 mU/l, range 16-1560; females: median 572 mU/l, range 20-3257) and in those not taking drugs capable of increasing serum PRL 363 mU/l (range 16-2565) (males: median 299 mU/l, range 16-1560; females: median 572 mU/l, range 20-2565). Serum PRL < 2000 mU/l was found in 98.7% (223/226) of the total group and in 99.5% (184/185) of those not taking drugs. Among the three subjects with serum PRL > 2000 mU/l, two were taking oestrogen preparations. Based on a large series of histologically confirmed cases, serum PRL > 2000 mU/l is almost never encountered in nonfunctioning pituitary macroadenomas. Values above this limit in the presence of a macroadenoma should not be surrounded by diagnostic uncertainty (after acromegaly or Cushing's disease have been excluded); a prolactinoma is the most likely diagnosis and a dopamine agonist should be considered as the treatment of choice.