Retrospective analysis on 312 patients, operated for thyroid nodules between 2014 and 2015, was conducted to evaluate the impact of frozen section analysis on the strategy of thyroid nodule surgery. One hundred and ninety-three patients were included. They all underwent preoperative US, fine needle aspiration cytology (FNAC), per operative frozen section (FS) and post operative definitive pathological analysis. Se, Sp, VPP and VPN of FNAC and FS were calculated and compared (McNemar's test). Multivariate analysis was performed to identify independent factor of good results. Se of FS and FNAC were, respectively, 86.1 and 81% with significant superiority of FS (p = .0352). Sp of FS and FNAC were, respectively, 100 and 72% with significant superiority of FS (p = .0156). A strategy based only on FNAC would have led to a 3.6% rate of unnecessary total thyroidectomy vs. 0% using FS. Overall rate of second procedure after lobectomy would have been significantly greater 28.9% without (28.9%) than with (10.3%) FS (p = .018). Overall rate of undone one-stage central neck dissection concurrent to total thyroidectomy for MNG would not have been significantly different without (9.4%) and with (2.1%) FS (.058). FNAC alone is unable to determine the extent of thyroid nodule surgery whatever the Bethesda subtype may be. FS significantly decreases the risk of two-stage procedure. For one-stage total thyroidectomy for MNG, the gain with FS is scarce.