There is a critical unmet need for new predictive and prognostic biomarkers for melanoma. This feature presents a brief overview of the opportunities and challenges facing investigators who are seeking biomarkers that can predict those melanomas that are likely to result in progression (metastasis) and death. Melanoma metastasis to the regional lymph nodes is now readily identified by sentinel node biopsy. However, currently available prognostic factors, which are based on clinical parameters and histological findings in the primary tumor, are limited in their ability to reliably discriminate which patients will manifest sentinel node metastasis and/or disseminated disease. To date, even those biomarkers for which substantial clinical correlation has been obtained have suffered from both conflicting data in clinical trials involving different patient populations and poorly understood interactions between the biomarkers and other established prognostic factors. An understanding of whether biomarker data should be viewed as a continuum or be categorized based on discrete breakpoints is also frequently lacking. In many cases, biomarkers are initially developed based on unequivocal index cases, such as clearly benign or clearly malignant melanocytic lesions, but then fail as predictive tools when applied to the more difficult cases that arise in clinical practice; such cases occupy a 'gray' zone that is filled with uncertainty and lacks an established 'gold standard' against which a biomarker prediction can be validated.