Immunoradiometric assay (IRMA) for serum prolactin (PRL) measurement can give falsely low values, leading to unnecessary surgery in patients with prolactinomas. We studied clinical and biochemical features of patients with pituitary macroprolactinomas in whom plasma PRL levels had been underestimated due to the so-called "high dose PRL hook effect". This phenomenon was observed in four (14.2%) out of 28 patients with pituitary macroadenomas (13 macroadenomas) and 15 non-functioning macroadenomas) reffer during one-year period. Undiluted median (range) PRL levels were 11.3 (3.0-48.7), 983.9 (194.4-1959.4), and 96.9 (66.6-147.7) micrograms/l in patients with non-functioning macroadenomas, macroprolactinomas and the hook effect adenomas, respectively. In all patients assay was performed after serum dilution, and only in patients with the hook effect the median PRL levels increased significantly to 5795.0 (2097.2-12722.2) micrograms/l. The mean age at diagnosis was 38 +/- 6.5, 45 +/- 6, and 53 +/- 3 yr, for the patients with the hook effect, macroprolactinoma and non-functioning adenoma, respectively. Males were predominant (75%) in the hook effect adenoma group. Patients with the hook effect macroprolactinomas were all treated successfully with dopamine agonists, and all patients had significant shrinkage of the tumor mass (more than 50% shrinkage). In conclusion, this study suggests that patients with high dose PRL hook effect are generally younger, more frequently males with very large pituitary adenomas (grade III-IV according to Hardy). It is necessary, whenever performing IRMA for serum prolactin measurement, to dilute samples routinely (1:1 and 1:10 dilutions) in every patient with pituitary tumor.