Melanoma is the most dangerous type of skin cancer and has highly heterogeneous features. Despite progress in melanoma classification, interpatient heterogeneity remains difficult to predict, especially in terms of long-term survival. Here, based on mRNA-seq, miRNA-seq and DNA methylation data from 447 cutaneous melanoma patients in the Cancer Genome Atlas, we performed integrative and single-dataset clustering analyses. A novel group of patients was identified, including 301 with better, 55 with poorer and 91 with intermediate prognoses. Immune genes were upregulated in the better prognostic group, and higher immune scores (representing a greater extent of immune cell infiltration into tumor tissues) were associated with better prognoses. Higher expression of 115 genes was determined to predict better outcomes. The better prognostic group also exhibited DNA hypomethylation, and immune pathways were enriched among the hypomethylated genes. Using exome-seq data from the same patients, we observed that the better prognostic group harbored the highest number of mutations. The mutational signature in the better prognostic group was associated with ultraviolet light exposure. These integrated investigations have potential therapeutic significance, as they clarify the molecular heterogeneity of cutaneous melanoma and enhance its classification.