Acquisition of an invasive phenotype of cancer cells in primary tumors is an absolute requirement for bone metastasis. The majority of bone metastases is derived from epithelial cancers, particularly those of the breast and prostate. Accumulating evidence suggest that transformed epithelial cells can activate embryonic programs of epithelial plasticity and switch from a sessile, epithelial phenotype to a motile, mesenchymal phenotype also referred to as epithelial-to-mesenchymal transition (EMT). Induction of EMT can, therefore, lead to invasion of surrounding stroma, intravasation, dissemination and colonization of distant sites. In bone/bone marrow disseminated tumor cells can partially regain their original epithelial characteristics via a mesenchymal-to-epithelial transition (MET) as glandular structures in bone metastasis are frequently observed. To date, the importance of epithelial plasticity in cancer cells disseminated to the bone/bone marrow microenvironment has remained largely elusive. Interestingly, a number of growth factors that play a prominent role in EMT induction in the primary tumor have been identified as important stimulators of skeletal metastasis formation. Recent studies have demonstrated that EMT may render cancer cells with properties of stem cells, which in turn can lead to escape from immune surveillance, increased resistance to apoptosis, diminished senescence and, last-but-not least, therapy resistance. This review will discuss current concepts regarding the role of epithelial plasticity in the multistep processes of bone metastasis, the issue of minimal residual disease, cancer stem cells and the importance of EMT in the development of novel targeted drug therapy.