The accuracy of diagnosis of 455 thyroid nodules evaluated by cytological and histological specimens obtained by fine- and large-needle biopsy, respectively, was compared with that for 1,094 nodules previously evaluated by clinical methods. The use of cytohistological data halved the number of patients with suspected cancer and doubled the number of patients to be observed. Cancers identified at operation for high and intermediate cancer-risk patients increased 75%, and operation for diagnosis of benign disease decreased 70%. Forty-two of 47 excised cancers were included in the cytohistological probable cancer group. Only 29 of these cancers were so classified clinically. Two cancers not diagnosed cytologically were suspected histologically and vice versa for one cancer. All 51 excised cytohistologically benign nodules were benign. There were more false-positive findings with cytological than with histological specimens.