Background: Long-term, randomized trial results comparing completion lymph node dissection (CLND) with observation for patients with sentinel lymph node (SLN) metastases are not available. Our goal was to determine whether melanoma patients with SLN metastases should undergo CLND. Methods: We developed a Markov model to simulate the prognosis of hypothetical cohorts of patients with SLN metastases who underwent either immediate CLND or observation with delayed CLND if macroscopic disease developed. Model parameters were derived from published studies and included the likelihood of non-SLN metastases, risk of dying from melanoma, CLND complication rates, and health-related quality-of-life weights. Outcomes included 5-year overall survival (OS), life expectancy (LE), and quality-adjusted life expectancy (QALE). Results: The projected 5-year OS for 50-year-old patients with SLN metastases who underwent immediate CLND was 67.2 % compared with 63.1 % for the observation group. The LE gained by undergoing immediate CLND ranged from 2.19 years for patients aged 30 to 0.64 years for patients aged 70 years. The QALE gained by undergoing immediate CLND ranged from 1.39 quality-adjusted life years for patients aged 30 to 0.36 for patients aged 70 years. In sensitivity analysis over a clinically plausible range of values for each input parameter, immediate CLND was no longer beneficial when the rate of long-term complications increased and the quality-of-life weight for long-term complications decreased. Conclusions: Immediate CLND following positive SLN biopsy was associated with OS and QALE gains compared with observation and delayed CLND for those who develop clinically apparent LN metastases.