A multicenter study evaluated the extent of observer variation in interpretation of hypocycloidal polytomograms of the sella turcica in women with amenorrhea and/or galactorrhea, some of whom had surgically verified pituitary adenomas. Three physicians—2 neuroradiologists and 1 neurosurgeon—independently and without additional clinical information about the patients reviewed 45 sets of tomograms, 16 of which had originally been classified as abnormal, 23 as normal, and 6 as equivocal. Two or more of these physicians agreed with the original reading in 58% of all cases: in 81% of those originally read as abnormal, 56% of those originally read as normal, but in none of those originally read as equivocal. In 8 women with surgically confirmed pituitary abnormalities, readers' accuracy ranged from 50 to 88%. The physicians varied considerably in the proportion of tomograms read as normal (16 to 45%) or abnormal (38 to 71%), but overall, for only 11% of the tomograms was there no agreement among any of the 3 reviewers. The findings indicate that even among highly experienced physicians, there is considerable variation in the interpretation of polytomograms of the sella for women with amenorrhea and/or galactorrhea.