The purpose of this study was to assess the risk of vocal palsy after thyroidectomy with identification of recurrent laryngeal nerve (RLN) during surgery. In all, 521 patients treated by the same surgeon were enrolled in this study. Temporary and permanent vocal palsy rates were analyzed for patient groups classified according to surgery for primary benign thyroid disease, thyroid cancer, Graves' disease, and reoperation. Measurement of the vocal palsy rate was based on the number of nerves at risk. Twenty-six intentionally sacrificed RLNs were excluded from analysis. Forty patients developed postoperative unilateral vocal palsy. Complete recovery of vocal palsy was documented for 35 of the 37 patients (94.6%) whose RLN integrity had been ensured intraoperatively. Recovery from temporary vocal palsy ranged from 3 days to 4 months (mean, 30.7 days). The overall incidences of temporary and permanent vocal palsy were 5.1% and 0.9%, respectively. The rates of temporary/permanent vocal palsy in groups classified according to underlying disease were 4.0%/0.2% for benign thyroid disease, 2.0%/0.7% for thyroid cancer, 12.0%/1.1% for Graves' disease, and 10.8%/8.1% for reoperation. Surgery for thyroid cancer, Graves' disease, and recurrent goiter were associated with significantly higher vocal palsy rates. Most patients without documented nerve damage during the operation recovered from postoperative vocal palsy. Total lobectomy with routine RLN identification is recommended as a basic procedure in thyroid surgery.