As melanoma cells are present in the circulation of many patients with this cancer, decreases in their number could provide an early indication of therapy effectiveness. To evaluate this possibility, we examined the effect of treatment with a melanoma vaccine on the number of melanoma cells present in the circulation. PCR was used to detect melanoma cells that expressed the melanoma-associated antigens MART-1, MAGE-3, tyrosinase and/or gp100 in 91 patients with melanoma. Melanoma cells that expressed one or more of these markers were present more often in advanced disease, i.e. in 80% of patients with advanced stage IV compared to in less than one-third of patients with less advanced disease. We then measured circulating melanoma cells in a subset of 43 of these patients who were treated with a polyvalent, shed antigen, melanoma vaccine. The vaccine contains multiple melanoma-associated antigens including MART-1, MAGE-3, tyrosinase and gp100. Immunizations were given intradermally q2-3 weeks x4 and then monthly x3. Prior to vaccine treatment, circulating melanoma cells were detected in 14 (32%) patients. Following 4 and 7 months of vaccine treatment, melanoma cells that expressed any of these markers were present in only nine (21%) and three (7%) of patients, respectively. Thus, vaccine therapy was associated with clearance of melanoma cells from the circulation in 78% of initially positive patients. As the number of these cells declined steadily with increasing length of therapy, it is unlikely that this was due to a random change in their number. Rather it suggests that the decline was a result of the therapy. These observations suggest that the presence of melanoma cells in the circulation is related to the extent of the melanoma, and that their disappearance may provide an early marker of the efficacy of therapy. However, the practical utility of assaying circulating tumor cells as a guide to the effectiveness of therapy or of prognosis will need to be confirmed by correlations with clinical outcome.