Several strategies have been developed to facilitate the prospective isolation of bone marrow-derived mesenchymal stem/stromal cells (BM-MSCs) based on the selective expression or absence of surface markers. Recently, we described the monoclonal antibodies W3D5 and W5C5, which selectively react with BM-MSCs, but not with hematopoietic cells. Both antibodies showed an identical reactivity pattern, indicating that they may recognize the same molecule. To identify the cognate antigen, cultured MSCs were sorted for cells expressing either very high levels of W5C5/W3D5 antigen or for cells which were negative for this antigen. Further processing of these cells for microarray analysis revealed a 20-fold enrichment of the type 1 integral membrane protein Sushi domain containing 2 (SUSD2) in the in W5C5(+) subset. To confirm the identity of the W5C5/W3D5 antigen to SUSD2, HEK293 cells were transfected with the full-length coding sequence of human SUSD2 followed by reactivity analysis of W5C5 and W3D5 antibodies with the transfected line. Flow cytometric analysis showed that both antibodies selectively recognized HEK293/huSUSD2 cells, but not the parental cell line. In line with this, SUSD2 siRNA treatment of SUSD2(+) WERI-RB-1 retinoblastoma cells reduced the expression levels of W3D5 and W5C5 antigens to ~39% and 37%, respectively. Finally, FACSorting and colony assays revealed that only SUSD2(+), but not SUSD2(-) BM cells give rise to colony-forming units-fibroblasts and are able to differentiate into osteoblasts, adipocytes, and chondrocytes. In conclusion, we identified SUSD2 as a novel and specific marker for the prospective isolation of BM-MSCs.