The invention as disclosed herein in encompasses a method for predicting the risk of metastasis of a primary cutaneous melanoma tumor, the method encompassing measuring the gene-expression levels of at least eight genes selected from a specific gene set in a sample taken from the primary cutaneous melanoma tumor; determining a gene-expression profile signature from the gene expression levels of the at least eight genes; comparing the gene-expression profile to the gene-expression profile of a predictive training set; and providing an indication as to whether the primary cutaneous melanoma tumor is a certain class of metastasis or treatment risk when the gene expression profile indicates that expression levels of at least eight genes are altered in a predictive manner as compared to the gene expression profile of the predictive training set.