SUMMARY. Lymph node involvement may impact postoperative therapeutic decision-making and prognosis in patients undergoing esophagectomy. This study evaluates which surgical approach yields the most lymph nodes. We undertook a retrospective chart review of esophagectomies performed by six surgeons from April 1994 to February 2004 using a prospective general thoracic surgery database at Mayo Clinic, Rochester, Minnesota, US. Lymph nodes were categorized into one of 17 regions per the American Joint Committee on Cancer, with the total number of lymph nodes, summed over each region, used as the primary outcome. A total of 517 esophagectomies were performed: 68 transhiatal, 392 Ivor Lewis, and 57 extended Ivor Lewis. A mean of 18.7 (SD 8.5) lymph nodes were retrieved with the Ivor Lewis approach as compared to 17.4 (SD 9.2) with the extended Ivor Lewis approach (P = 0.30). Since there was no statistical difference between the number of nodes collected in either Ivor Lewis approach, they were collapsed into one group for comparison with the transhiatal cases. Significantly more lymph nodes were collected with an Ivor Lewis approach (mean 18.5, SD 8.6) than with a transhiatal approach (mean 9.0, SD 5.0, P < 0.001). As expected, more thoracic lymph nodes were retrieved with the Ivor Lewis approach [mean 12.4 (SD 7.0) vs. 4.7 (SD 5.3), P < 0.001]. The Ivor Lewis approach was also superior for retrieval of abdominal nodes [mean 6.1 (SD 5.6) versus 4.3 (SD 4.4), P = 0.01]. More lymph nodes are obtained at esophagectomy with an Ivor Lewis than a transhiatal approach.