BACKGROUND: Skin melanoma exhibits significant heterogeneity in clinical outcomes and treatment responses among patients. This study aimed to investigate natural killer (NK) cell clusters in skin melanoma, their impact on patient prognosis, and their value as biomarkers for tailoring treatment. METHODS: We used data from TCGA, GSE19234, GSE65904, GSE244982, and GSE78220. A gene classifier was developed to identify two distinct clusters of melanoma patients. Survival analysis, NK cell infiltration levels, and responses to immune and targeted therapies were evaluated. RESULTS: Unsupervised clustering revealed two distinct melanoma patient clusters with significant differences in NK cell activity and clinical outcomes. Cluster 1 showed higher NK cell infiltration, better overall survival (OS) (p <
0.0001), and greater activity in NK-cell-related pathways. In contrast, Cluster 2, characterized by lower NK cell activity and higher exhaustion markers, had poorer OS. Drug sensitivity analysis indicated that Cluster 1 was more responsive to most melanoma treatments, whereas Cluster 2 had higher sensitivity to trametinib (p <
0.001). The developed gene classifier had an AUC of 0.913 and effectively differentiated between clusters. Additionally, Cluster 1 showed better responses to immunotherapy with a higher rate of complete and partial responses (p <
0.001). These findings were validated in external databases. CONCLUSION: This study identifies two distinct NK-cell-related clusters in melanoma with differential prognoses and treatment responses. These findings underscore the importance of integrating NK-cell-related profiles into personalized treatment strategies, offering a pathway to optimize therapeutic outcomes based on specific immune profiles.