Tumorigenesis in humans and laboratory animals is a com plex, mu11ÂĄstepprocess (1, 2). In humans, in whom the process has been studied only indirectly, measurements of age-depen dent tumor incidence indicate kinetics dependent on the fifth or sixth power of elapsed time (3). This suggests a succession of five or six independent steps, each of which is rate limiting on the process. In experimental models, such as mouse skin tumorigenesis, the process has been broken down into at least three distinct steps: initiation, promotion, and progression (4, 5). From the perspective of the organism, the multistep nature of tumorigenesis is easily rationalized; each step in the process represents a physiological barrier that must be breached in order for a cell to progress further toward the end point of malignancy. Such multiple barriers conspire to ensure that successful completion of the tumorigenic process is a rarely achieved event.

An unanswered question concerns the natures of these bar riers to tumor inception and growth. A portion of the defenses may well derive from systemic defenses against tumors; yet others, confronted in this essay, reflect underlying mechanisms governing the behavior of individual cells. Our cells and likely those of all metazoa would seem to be constructed so as to present multiple impediments to full malignant transformation. Only recently has it been possible to search for the molecular and cellular mechanisms that govern multistep tumorigenesis. What are the rules that govern cell growth? How is the growthregulatory circuitry laid out within the cell? And how can multiple physiological controls be overridden to produce the deregulation of neoplasia?