Methods have recently been described for purification of mesenchymal stem cells (MSCs) from human bone marrow (BM), based on different combinations of antibodies and magnetic beads (Quirici et al, 2002, Reyes et al, 2002, Jones et al, 2002, Gronthos et al, 2003). We have previously demonstrated that BM MSCs reside in the CD45lowD7-FIB+LNGFR+ cell fraction (Jones et al, 2002). In the present work we compared four different microbead-based enrichment methods and showed that positive selection with anti-fibroblast (D7-FIB) microbeads (n=28), LNGFR microbeads (n=5) and plastic adherence/trypsinization (n=6) resulted in similar enrichment of cells with this phenotype, with their frequency rising from 0.01% in un-fractionated marrow to 6.6±5.8%, 7.7±7.0 and 4.9±1.5%, respectively, in the enriched fraction. Although MSC isolation by negative depletion with CD45 and GPA microbeads produced poorer enrichment (1.3±0.9%, n=5) its advantage was that MSCs remained unlabelled. Defining the best positive marker for selection requires its high expression on MSCs (high specificity) and lack of binding to CD45+ cells (high selectivity). LNGFR remained the best marker, followed by CD146, CD106, D7-FIB, CD13 and CD166. In contrast, CD105, CD73, CD44 and CD63 were not very selective and showed only moderate levels of expression on MSCs. Mesenchymal nature of sorted CD45lowLNGFR+ cells was further confirmed by real-time PCR measurements of transcription factors (TFs) implicated in mesenchymal lineage differentiation towards bone, cartilage and adipose tissue. Compared with matched haematopoietic lineage cells (CD45+ cells) freshly purified CD45lowLNGFR+ MSCs showed significantly higher expression of cbfa-1, Sox9 and PPARγ ranging from 20- to 150-fold. IL-7 RNA was also 60-fold more abundant in sorted MSCs than in CD45+ cells. Whether these TFs and IL-7 were expressed by the same cells or in separate populations of osteoprogenitors and preadipocytes, remains to be established. Oil red staining of individual, freshly sorted MSCs, however, revealed that some cells contained microscopic fat droplets and thus were already engaged in adipogenic progression, particularly in older individuals. Sox9 and cbfa-1 expression levels were similar in freshly purified and expanded MSCs, whilst the expression of PPARγ and IL-7 declined during standard expansion, but could be restored in adipogenic and stromal media, respectively. In vivo levels of stromal differentiation and adipogenic progression therefore appear to be transiently lost following standard MSC expansion. Whilst adipogenesis seems to be environmentally regulated, osteogenesis, on the other hand, may indeed be a default state of MSCs in the BM. Consistent with this idea, osteogenesis-related markers (alkaline phospatase, and BMP receptors) were uniformly, and not differentially, expressed by freshly sorted MSCs. Finally, freshly purified MSCs did not express ES-cell markers SSEA-3, SSEA-4, and Tra-160, SP-population specific protein ABCG2 and flk-2/flt3, but were positive for Tra-181. Altogether, these results not only consolidate previous independent reports on the phenotype and purification of in vivo human BM MSC, but also shed light on their state of differentiation and relationship with other adult stem cells.