Patterns of reproductive uncertainty can have an important influence on population dynamics. There is a crucial distinction between what we describe here as aggregate uncertainty (in which reproductive output in each generation is correlated among the individuals in a population) and idiosyncratic risk (in which reproductive output is independent across individuals). All else being equal, populations experiencing idiosyncratic risk enjoy a higher asymptotic growth rate than do those experiencing aggregate uncertainty. Therefore individuals in populations of the former type will have a competitive advantage over individuals in populations of the latter type. Applying this distinction to models of randomly fluctuating environments, we point out that genetic variation among offspring can serve to reduce aggregate uncertainty, transforming it into a more idiosyncratic form of risk. We show that this transformation underlies the dynamics observed in several previous models of the role of outcrossing in the evolution of sex.