BACKGROUND: The objectives of this analysis were to compare various measures associated with lymph node (LN) dissection and to identify threshold values associated with disease-specific survival (DSS) outcomes in patients with melanoma. METHODS: Patients with lymph node-positive melanoma who underwent therapeutic LN dissection of the neck, axilla, and inguinal region were identified from the SEER database (1988-2005). We performed Cox multivariate analyses to determine the impact of the total number of LNs removed, number of negative LNs removed, and LN ratio on DSS. Multivariate cut-point analyses were conducted for each anatomic region to identify the threshold values associated with the largest improvement in DSS. RESULTS: The LN ratio was significantly associated with DSS for all LN regions. The LN ratio thresholds resulting in the greatest difference in 5-year DSS were .07, .13, and .18 for neck, axillary, and inguinal regions, respectively, corresponding to 15, 8, and 6 LNs removed per positive lymph node. After adjustment for other clinicopathologic factors, the hazard ratios (HRs) were .53 (95% confidence interval [CI], .40 to .71) in the neck, .52 (95% CI, .42 to .65) in the axillary, and .47 (95% CI, .36 to .61) in the inguinal regions for patients who met the LN ratio threshold. CONCLUSIONS: Among the prognostic factors examined, LN ratio was the best indicator of the extent of LN dissection, regardless of anatomic nodal region. These data provide evidence-based guidelines for defining adequate LN dissections in melanoma patients. (c) 2009 American Cancer Society.