As diverse therapeutic options are now available for advanced melanoma patients, predictive markers that may assist treatment decision are needed. A model based on baseline serum lactate dehydrogenase (LDH), peripheral blood relative lymphocyte counts (RLC) and eosinophil counts (REC) and pattern of distant metastasis, has been recently proposed for pembrolizumab-treated patients. Here, we applied this model to advanced melanoma patients receiving chemotherapy (n =116) or anti-CTLA-4 therapy (n =128). Visceral involvement, LDH and RLC were associated with prognosis regardless of treatment. Instead, when compared to chemotherapy-treated patients with REC < 1.5%, those with REC ≥ 1.5% had improved overall survival when receiving anti-CTLA-4 [Hazard Ratio (HR) = 0.56 (0.4-0.93)] but not chemotherapy [HR = 1.13, (0.74-1.74)], and the treatment-by-REC interaction was significant for both overall (p= 0.04) and progression free survival (p= 0.009). These results indicate baseline REC ≥ 1.5% as a candidate predictive biomarker for benefit from anti-CTLA-4. Further studies are needed to confirm these findings in patients receiving immune-modulating agents.