Résumé : The aim of the present work was to investigate the origin and half-life of endogenous S100B protein reported by many investigators as a useful melanoma serum marker. Within cells, S100B protein exists in homo- or heterodimer form containing mainly Ca(++), having a substantial fraction bound to membranes. As such, S100B is believed to be involved in the regulation of cytoskeleton. Also, a role in the cell cycle progression has been suggested. Although S100B appears having important intracellular functions, proofs of its secretion, at least at concentrations such as the ones measured in melanoma patients, are still lacking. Consistent with this view is the fact that immunohistology for S100 protein reported by numerous authors clearly indicate an exclusive intracellular staining. For these reasons, it was of a major interest to investigate how and when S100B is shed to the blood. Knowing that significant S100B levels are seen only in stage IV patients, we hypothesized that cell death may be the major source of circulating S100B protein in these patients. This hypothesis was studied in an in vitro model simulating cell death and in vivo in melanoma and other cancer patients undergoing highly cytotoxic regional immunochemotherapy using isolated limb perfusion with tumor necrosis factor and melphalan, as well as in tumor exudates and pleural fluids. Our results strongly suggest melanoma and endothelial cell death and subsequent continuous drainage to the blood as the major mechanism behind S100B release to the blood circulation. We estimated the endogenous S100B protein half-life to be about 30 min.