Background: Because fine-needle aspiration cannot reliably discriminate between benign and malignant follicular thyroid lesions, some surgeons use intraoperative frozen section (FS) to guide operative management. To determine the utility of FS for these lesions, we reviewed our institutional experience. Methods: Between 1994 and 2001, 152 patients underwent surgical resection for follicular neoplasms. Results: The mean age of the patients was 47 years, and 76% were female. Forty-one (32%) FSs were reported as benign, 5 (4%) as malignant, and 3 (2%) as indeterminate, and in 80 (62%), the diagnosis was "follicular lesion, deferred to permanent histology." On paraffin section, all patients with malignant FSs had thyroid cancer, and all 41 patients with benign FSs had benign lesions. Thus, FS for diagnosis of follicular thyroid cancer had a sensitivity, specificity, positive predictive value, and accuracy of 67%, 100%, 100%, and 96%, respectively. In most cases (64%), FS rendered no additional information at the time of operation. Therefore, the cost per useful FS was $7800, which is higher than the charge of a completion thyroidectomy (approximately $6000). Conclusions: FS analysis for follicular lesions seems to be highly specific and accurate. However, because of the low sensitivity, routine use of FS is not cost-effective in patients with follicular thyroid lesions. © 2003 The Society of Surgical Oncology, Inc.