Background: Sentinel lymph node (SLN) biopsy for melanoma often detects minimal nodal tumor burden. Although all node-positive patients are considered stage III, there is controversy regarding the necessity of adjuvant therapy for all patients with tumor-positive SLN. Methods: Post hoc analysis was performed of a prospective multi-institutional study of patients with melanoma ≥ 1.0 mm Breslow thickness. All patients underwent SLN biopsy; completion lymphadenectomy was performed for patients with SLN metastasis. Kaplan-Meier analysis of disease-free survival (DFS) and overall survival (OS) was performed. Univariate and multivariate Cox regression analyses were performed. Classification and regression tree (CART) analysis also was performed. Results: A total of 509 patients with tumor-positive SLN were evaluated. Independent risk factors for worse OS included thickness, age, gender, presence of ulceration, and tumor-positive non-SLN (nodal metastasis found on completion lymphadenectomy). As the number of tumor-positive SLN and the total number of tumor-positive nodes (SLN and non-SLN) increased, DFS and OS worsened on Kaplan-Meier analysis. On CART analysis, the 5-year OS rates ranged from 84.9 % (women with thickness < 2.1 mm, age < 59 years, no ulceration, and tumor-negative non-SLN) to 14.3 % (men with thickness ≥ 2.1 mm, age ≥ 59 years, ulceration present, and tumor-positive non-SLN). Six distinct subgroups were identified with 5-year OS in excess of 70 %. Conclusions: Stage III melanoma in the era of SLN is associated with a very wide range of prognosis. CART analysis of prognostic factors allows discrimination of low-risk subgroups for which adjuvant therapy may not be warranted. © 2012 Society of Surgical Oncology.