Observations have been reported which indicate that mice inoculated intracerebrally with active yellow fever virus may develop an infection which is not only non-fatal but may also be completely inapparent. The most extensive observations were made on mice which showed signs of infection but were still alive 22 days after inoculation with virus of one or another of several 17D substrains. In such cases, the infection usually progressed no further and partial or complete recovery often ensued. Agents other than yellow fever virus were excluded as a significant cause of such nonfatal infections by the failure of repeated attempts to isolate other infective agents, by the demonstration of antibodies against yellow fever virus in the sera of the mice, and by the demonstration of a high degree of resistance on the part of such surviving mice to reinoculation with large doses of neurotropic yellow fever virus. Completely inapparent infections with 17D virus were also shown to occur. Studies of apparently normal survivors of 17D virus titrations revealed a small but significant number of animals resistant to intracerebral challenge with neurotropic yellow fever virus. Further, pooled sera from such mice were shown to contain specific protective antibodies. The occurrence of non-fatal infections with 17D virus was found related to virus dose and substrain. Small doses of virus provoked a significantly higher proportion of non-fatal infections than large doses; while different 17D substrains, tested over equivalent ranges of virus dose, varied greatly with respect to the proportion of infections which did not terminate with death. In the case of two substrains (17DD low and 17D(3)), non-fatal infections (as demonstrated by resistance to intracerebral challenge with neurotropic virus) were sufficiently frequent to cause an increase, when included in the computation of the infective titers, of 25 per cent above the figures based on deaths alone. The demonstration of non-fatal infections, thus, may be important to the accuracy of quantitative determinations of infectivity. Limited observations with virus of the French neurotropic and the pantropic Asibi strains revealed that non-fatal infections do occur, but only rarely. Somewhat more extensive observations with unmodified virus of strains isolated from Brazilian cases of jungle yellow fever, in contrast, revealed an occurrence of non-fatal infections much greater than that observed with the most productive 17D substrains. With these jungle strains, the demonstration of non-fatal infections proved indispensable to any measure of the level of infectivity of virus preparations. The demonstration of the proportional occurrence in mice of non-fatal infections with yellow fever virus provides an additional means by which different virus strains and substrains may be characterized.