Regional lymph nodes are the first site for melanoma metastases. The sentinel node (SN), on the direct lymphatic drainage pathway, which usually harbors first metastases, demonstrates significant suppression in its ability to respond to antigenic stimulation. This down-regulation of SN immunity is likely the basis of its susceptibility to tumor metastases, suggesting a potential role of the immune system in the control of malignant tumors. Despite immune dysfunction in the SN, phase II trials of systemic post-operative immunotherapy with a polyvalent melanoma vaccine developed at the John Wayne Cancer Institute showed improved 5-year overall survival in patients with melanoma metastatic to regional nodes. However, most immunotherapy clinical trials have failed to demonstrate a significant clinical response, and analyses of immune responses to tumor-associated antigens that correlate clinical responses have not been established. Therefore, refinements in assay methodologies and improvements in vaccine designs are critical to the success of cancer immunotherapy. Antigen presentation by dendritic cells (DCs) is the most potent means to initiate a T cell immunity. Dendritic cell-based immunotherapies have been vigorously attempted in the past decade. To improve the immunogenicity of cancer vaccines, we recently generated heterokaryons of DCs and tumor cells by electrofusion. The fusion hybrids retained their full antigen-presenting capacity and all natural tumor antigens. In pre-clinical animal experiments, a single injection of the DC-tumor fusion hybrids was sufficient to mediate the regression of tumors established in the lung, skin and brain. Most interestingly, successful therapy required the delivery of fusion hybrids directly into lymphoid organs such as lymph nodes. A clinical trial is now being carried out to test the immunogenicity and therapeutic effects of fusion hybrids for the treatment of metastatic melanoma.