The studies described above have led to some conclusions as well as some speculations regarding the participation of oncogenes in proliferation. It is important to make a clear distinction between the data described along with the resulting conclusions, and the highly speculative models which have been proposed here to describe these results. On the basis of the results described, several conclusions appear to be indicated. First, it is clear that in normal cells of many types ras proteins are required for proliferation. Second, these proteins are needed for the maintenance of the transformed phenotype induced by some, but not all, oncogenes. Third, the activity of ras proteins is not apparently involved in controlling the activity of phospholipase C or any other identifiable phospholipase. Finally, tumor formation appears to involve the development of a proliferative phenotype which functions independently of ras activity. To explain these data, a model of proliferative control is presented. This model is highly speculative at this time and is based upon the assumption that the "ras pathway" involves the sequential action of cellular genes related to the retroviral oncogenes. The function of this pathway is to pass the proliferative signal generated by an occupied growth factor receptor into the cell. It is clearly needed late in the G-1 phase of the cell cycle, but the "ras pathway" might also be involved in the early events associated with mitogenic stimulation. Due to its central role in the control of proliferation and the fact that tumor cells circumvent its action, ras proteins are postulated to a site of negative proliferative control. After the action of ras and related cellular oncogenes, the proliferative control signal no longer involves the action of a simple linear sequence of protein activities, but might involve multiple, interdependent pathways. This model is primarily of value as a working hypothesis and does not account for many observations central to proliferative control such as the involvement of cell-cell contact, differentiation, and the action of factors which inhibit rather than promote proliferation, such as interferons. The model does summarize the data described and even in the simplest form represents a novel approach to explain proliferative control in terms of the activities of known genes.