A variety of immunotherapeutic approaches have shown activity in patients with metastatic melanoma with the best results being observed with interleukin 2 (IL-2). Follow-up data through 2004 confirm the durability of responses produced by the Food and Drug Administration-approved high-dose IL-2 regimen in this patient population. Efforts to develop more tolerable and/or effective IL-2-based treatment regimens by either prolonged administration of lower doses or the combination of IL-2 with other cytokines, monoclonal antibodies, or vaccines have yet to produce results superior to those seen with high-dose IL-2 alone. Recent investigations have suggested that, in some patients, IL-2 may expand regulatory T-cell populations leading to immune tolerance rather than antitumor immunity. Efforts to shift this balance in favor of immune rejection by reducing the confounding effects of regulatory T cells on IL-2 therapy or the use of novel and potentially more purely immunostimulatory cytokines are ongoing. Despite promising phase 2 data, phase 3 studies have failed to show meaningful clinical benefit for the combination of cytokines with cytotoxic chemotherapy, so-called "biochemotherapy." Nonetheless, recent investigations with biochemotherapy followed by maintenance immunotherapy suggest that biochemotherapy may still have a role as a "bridge to immunotherapy" in some patients with rapidly progressive disease. Given the low number of patients achieving durable benefit with cytokine-based immunotherapy, considerable recent effort has focused on identifying predictors of therapeutic response. Investigations suggest that immune responsiveness may be predetermined by a tumor microenvironment conducive to immune recognition and the host propensity to develop autoimmunity. Efforts to understand and further define pretreatment predictors of response through the use of gene expression and proteomic techniques are ongoing and raise the potential for eventually limiting cytokine-based immunotherapy to those most likely to benefit.