1. Leukemia, like other malignant tumors, has long been observed to occur more frequently among members of certain families than in the average population. This observation has been made not only in chickens, mice, and cattle, but also in humans. 2. Direct experimental evidence is now available suggesting that at least in chickens and mice, leukemia is caused by a filterable, submicroscopic agent, presumably a virus, transmitted from one generation to another directly through the embryos. 3. Although no experimental evidence is yet available concerning the nature of leukemia in cattle and in humans, the possibility should be considered that, as in chickens and mice, leukemia in other species may also be caused by a transmissible agent. In cattle, numerous observations have been made of leukemia developing in several members of the same family, in two or three successive generations. In humans, one case has been published in which three, and perhaps four, out of seven, and another in which five out of eight, siblings died from leukemia ; other observers have recorded leukemia developing in three successive generations in man. 4. The familial occurrence of leukemia and other malignant tumors in man and other species, could be explained by the following working hypothesis: there may exist a group of submicroscopic, cell-free oncogenic agents, individually distinct (some causing leukemia, others causing different malignant tumors), transmitted directly through the germinal cells from one generation to another. These agents would usually exist in an inactive form; in such a form, these agents would be frugal and moderate in their requirements, causing no apparent harm to their carrier hosts. Occasionally, however, prompted by as yet obscure but presumably varied trigger stimuli, these hitherto masked agents might change into formidable pathogens. Activated oncogenic agents would cause rapid multiplication of cells harboring them, resulting in the development of leukemia or other malignant new growth, and usually sealing the fate of their carrier hosts. Since, however, the activation of the oncogenic agent would usually occur in middle-aged hosts, the survival of the agent would have been already assured, the transmission of the agent to the host’s offspring occurring prior to the activation of the agent. Leukemia and other malignant tumors would therefore be essentially egg-borne virus diseases, like many other parasitic virus diseases of plants, insects, and birds. 5. On the basis of this working hypothesis, one could therefore regard the development of leukemia, or that of another malignant tumor, as the result of activation, frequently merely accidental, of an oncogenic agent, hitherto, masked, and carried by the host since birth. Since in many, perhaps in most, instances, the oncogenic agents would never become activated (though they would continue being transmitted from one generation to another), only now and then would leukemia, or another malignant new growth, develop in some of the members of certain families. Such instances, however, would usually be separated by one, two, or three generations of healthy individuals. 6. Since, therefore, two cases of leukemia (or other malignant tumors) representing successive links in the generation-to-generation transmission of the hypothetic virus in humans might frequently be separated by a century or two, the tracing of the epidemic course of the oncogenic agents in man would be extremely difficult, if at all possible, unless accurate records of cause of death would become available covering a sufficient number of successive generations. 7. Should this working hypothesis prove to be true, it would then follow that the number of individuals suffering from leukemia, or other tumors, would represent only a fraction of those actually carrying the seeds of either disease.