Multicenter selective lymphadenectomy trial 1 (MSLT-I) defined the prognostic and potential therapeutic values of sentinel lymph node biopsy (SLNB) for intermediate-thickness melanoma. The role of completion lymphadenectomy (CLND) is, however, unclear and the subject of the ongoing MSLT-II trial. From 2003 to 2012, patients with tumors 1-4 mm thick with positive SLNB were identified in the Surveillance Epidemiology and End Results Program registry. The patients were divided into two groups: group 1 (CLND) and group 2 (observation). The study enrolled 2172 patients, the majority of whom were white and male with extremity primaries, no ulceration, Clark level 4 invasion, and nodes 2.01-4.0 mm deep. In the univariate analysis, CLND was associated with lower mean age, male gender, primary site, number of positive nodes, and geographic region (p<0.05). In the multivariate analysis, male gender [odds ratio (OR), 1.27] and geographic area (Michigan OR, 2.31; Iowa OR, 1.69) were associated with CLND (p<0.05). In the survival analysis, male gender, primary site, ulceration, Clark level, and depth and number of positive nodes were associated with survival (p<0.05), but CLND was not (p=0.83). In the Cox regression analysis, the relationship between male gender [hazard ratio (HR), 1.14], primary site trunk versus extremity (HR, 1.3), ulceration (HR, 1.79), Clark level (2 vs. 4 HR, 3.51; 2 vs. 5 HR, 6.48), depth (HR, 1.43) and number of nodes (1 vs. 2: HR, 1.23; 1 vs. ≥3: HR, 2.52) persisted (p<0.05). However, when CLND was included in this model, it was not associated with improved survival. Age, gender, and geographic area predict the likelihood of CLND. In this retrospective study, CLND did not add survival benefit.