Variance in reproductive success (Vk, with k = number of offspring)is the stuff of evolution. Evolution by natural selection requires variation in offspring number, and the ratio ratio phi = Vk/kbar is the primary factor that reduces effective size (Ne) compared to census size (N). To date, no attempts to parse factors that contribute to Vk have focused on age-specific values of phi, which indicate the degree to which reproductive skew is overdispersed among individuals of the same age and sex. Here, an ANOVA framework is used to partition Vk into between-age and within-age components and, for lifetime reproductive success, a component due to variation in longevity. The between-age effect depends on mean age-specific fecundity (bx), but few empirical data are available on the within-age effect, which depends on phi. Analytical results show how to quantify the expected contribution of random demographic stochasticity to these variance components. Analyses of empirical and simulated data illustrate the new methods and show that the largely-neglected within-age effect a) typically represents a substantial component of overall variance in offspring number (even under a null model of random reproductive success), and b) can dominate the overall variance when phi >1.