The learning curve for robotic thyroidectomy with central compartment node dissection (CCND) has not been established. We examined the effect of experience of robotic thyroidectomy on a range of perioperative parameters in order to determine the learning curve. The learner surgeon outcomes were compared with those of an experienced surgeon. We conducted a prospective, controlled, multicenter study involving four endocrine surgeons at three academic centers. Patients underwent robotic total or subtotal thyroidectomy with CCND between September 2008 and October 2009. One surgeon was experienced in the technique (experienced surgeon, ES), while the other three surgeons had endoscopic thyroid surgery experience but no experience performing the robotic procedure (nonrobotic thyroid surgery experienced surgeon, NS). Outcome measures were demographic data, operative time, blood loss, hospital stay, pathologic results, and postoperative complications. A total of 644 total or subtotal robotic thyroidectomies with CCND were performed: 377 (58.7%) by NSs and 267 (41.5%) by the ES. Mean operative time was longer and the complication rate was higher for the NS patient group compared with the ES patient group (P < 0.001 for each). The operative times and complications rates for the NS group were similar to those of the ES group once the NSs had performed 50 cases for total thyroidectomies or 40 cases for subtotal thyroidectomies. The learning curve duration for robotic thyroidectomy with CCND using gasless transaxillary approach for experienced endoscopic thyroidectomy surgeons was 50 cases for total thyroidectomy and 40 cases for subtotal thyroidectomy.