The aim of this study was to evaluate S100B in bone marrow (BM) plasma from malignant melanoma patients. BM aspirates and peripheral blood (PB) plasma from 56 patients and BM aspirates from 29 healthy volunteers were collected. S100B was measured using an immune radiometric assay, which is a two-site sandwich assay based on monoclonal antibodies recognizing the β-subunit. In the control population, the median S100B level in BM plasma was 9.0 μg/l (26 women and three men), an unexpectedly high value compared with the median S100B level in PB<0.05 μg/l. S100B levels in BM seems to be sex dependent. Median S100B levels in samples taken from male melanoma patients was 26.7 μg/l in contrast to 9.3 μg/l in female patients (Mann–Whitney P<0.002). The elevated BM S100B in melanoma patients could not be explained by presence of melanoma cells in the BM, as the values also were increased to the same extent in patients with no detectable BM metastases. In attempts to identify the source of S100B in BM, cytospins from five patients with high S100B values were stained, but none of the BM cells stained positive. S100B levels in PB were dependent on the stage of melanoma disease and there was a significant shorter survival time in the group of patients with elevated S100B compared with the group with normal S100B values, (log rank test: P=0.04). In BM taken from melanoma patients, however, there were no association between S100B levels and survival. The median S100B level in BM aspirates from healthy female volunteers and BM samples from female melanoma patients were 8.1 and 9.3 μg/l both manifold higher than the cut-off value for S100B in PB (0.2 μg/l). The median S100B in the samples taken from male melanoma patients was nearly three times higher than in the female patients. Unlike S100B in PB, S100B in BM demonstrated no prognostic value. The explanation for the unexpected high S100B in BM remains elusive.