SUMMARYMesenchymal stem cells (MSCs) have been identified in several adult tissues and have the ability to differentiate into multiple cell types. However, not all MSC types differentiate efficiently into all lineages. We show here that there is an epigenetic basis for this restricted differentiation capacity. DNA methylation was determined on lineage-specific promoters by bisulfite sequencing, and genome-wide by methyl DNA immunoprecipitation (MeDIP) with promoter array hybridization in mesenchymal stem cells (MSCs) of various tissues. Bisulfite sequencing shows that the adipogenic FABP4 and PPARG2 promoters are differentially methylated in ASCs and bone marrow (BM) MSCs relative to Wharton’s jelly (WJ) MSCs and muscle progenitor cells (MPCs). In contrast, the myogenin (MYOG) promoter is hypomethylated in WJMSCs and MPCs relative to ASCs and BMMSCs. In hematopoietic stem cells (HSCs), all promoters are hypermethylated. Differential methylation correlates with distinct differentiation capacities: ASCs and BMMSCs differentiate efficiently into adipocytes but not into multinucleated myogenin positive myocytes, whereas MPCs display poor adipogenic differentiation but robust myogenic capacity. The endothelial CD31 gene is methylated in ASCS, BMMSCs, WJMSCs and MPCs, in agreement with their poor endothelial differentiation potential. CD31 is however unmethylated in HSCs, in which its expression can be induced. Methylation patterns in adipocytes, muscle and endothelial cells argue that ASCs (and BMMSCs) are epigenetically pre-programmed for adipogenesis, while MPCs have a methylation pattern predictive of myogenic potential. Bisulfite sequencing corroborates genome-wide methylation profiling. MeDIP reveals similarity in methylation profiles between MSCs from adipose tissue, bone marrow and muscle, reflecting the mesodermal origin of these cells. Our results put forward the hypothesis that DNA methylation patterns on lineage-specific promoters may predict lineage differentiation capacity.