Background The extent of surgery for well‐differentiated thyroid cancer remains controversial. The purpose of this study was to evaluate the type of resection, age, T classification, nodal status, tumor size, and year of diagnosis for overall survival (OS) and cause‐specific survival (CSS) using a large database. Methods Using the Surveillance, Epidemiology, and End Results (SEER) database, 23,605 subjects were identified with papillary or follicular thyroid cancer between 1983 and 2002. OS and CSS were estimated, and outcomes for local excision, lobectomy, near‐total thyroidectomy, or total thyroidectomy were compared. Results Ten‐year OS and CSS by surgery were: total thyroidectomy, 90.4% and 96.8%, respectively; near‐total thyroidectomy, 89.5% and 96.6%, respectively; and lobectomy, 90.8% and 98.6%, respectively. Controlling for risk factors, near‐total thyroidectomy was inferior to total thyroidectomy for OS (hazard ratio HR 1.21; p = .019) and CSS (HR 1.39; p = .019). Age, T3/T4 disease, positive nodes, and tumor size were associated with poorer outcomes. Conclusion Total thyroidectomy resulted in improved survival. Therapy should be individualized, accounting for potential complications and recurrence patterns. © 2010 Wiley Periodicals, Inc. Head Neck, 2011