Human embryonic stem (ES) cells offer substantial opportunities for providing well-defined differentiated cells for drug discovery, toxicology, and regenerative medicine, but the development of efficient techniques for their large-scale culture under defined conditions, and for controlling and directing their differentiation, presents a substantial challenge. Markers for defining the undifferentiated cells are well established, based upon previous studies of embryonal carcinoma (EC) cells, their malignant counterparts from teratocarcinomas. These provide valuable tools for monitoring human ES cultures and their state of differentiation. However, current culture techniques are suboptimal and involve the use of poorly defined culture media and the use of feeder cells. Over time, the cells may also acquire karyotypic changes, reflecting genetic selection and adaptation to in vitro culture conditions. Nevertheless, progress is being made. Originally, human ES cells were derived and maintained in medium containing fetal calf serum. They are now widely cultured in a proprietary serum-free formulation (serum replacement from Invitrogen Corp., Carlsbad, CA), and recently we have derived a new human ES line in this medium without fetal calf serum. Human fibroblasts can also be used to replace mouse embryo fibroblasts as feeder cells. We have now found it possible to culture a subline of human ES cells on Matrigel, or purified collagen type IV, laminin, and fibronectin, without feeders or feeder-conditioned medium. These cells nevertheless retain the features of undifferentiated human ES cells, including a capacity for differentiation. Although these cells also carried karyotypic changes, further research focused upon understanding the mechanisms that control self-renewal, apoptosis, and commitment to differentiation will facilitate the development of defined culture conditions that minimize genetic change and optimize the maintenance of the undifferentiated stem cells.