Studies of bone cells in culture have raised two salient questions: are the findings representative of the in vivo situation and can the conflicting data from different cell models be reconciled? Review of the literature indicates that all osteoblastic cells, defined by their origin or by their ability to produce mineralized matrix, have a few common properties: production of type I collagen; increased alkaline phosphatase activity; and parathyroid hormone-stimulated adenylate cyclase. Other features, such as osteocalcin and prostaglandin E production and the response to prostaglandin E, are selectively expressed by certain cell types. Pilot studies on mRNA levels of 'bone proteins' in developing calvaria suggest that such differences may reflect stages in osteoblastic differentiation. Immortalization of calvaria-derived cells using a SV40 large T antigen vector, which may freeze the cells in their particular state of differentiation (as proposed for leukaemia cells), yields phenotypes consistent with that hypothesis. Immortal cell lines may thus help to characterize osteoblastic differentiation. The diversity of osteoblast responses in culture to hormones and growth factors could be due to these phenotype differences but could also represent a subspecialization of differentiated cells. In addition, in the organism regulatory agents act in concert on a heterogeneous interactive cell population. Nonetheless cell cultures can be useful in screening for and predicting in vivo responses, as was shown by the 1,25-(OH)2D3 stimulation of osteocalcin, and for studying the molecular mechanisms of regulatory effects. Cell lines are also convenient for the production of specific proteins and cDNA libraries, and for the expression of specific genes.