Most models of optimal progeny size assume that there is a trade-off between progeny size and number, and that progeny fitness increases with increasing investment per young. We find that both assumptions are supported by empirical studies but that the trade-off is less apparent when organisms are iteroparous, use adult-acquired resources for reproduction, or provide parental care. We then review patterns of variation in progeny size among species, among populations within species, among individuals within populations, and among progeny produced by a single female. We argue that much of the variation in progeny size among species, and among populations within species, is likely due to variation in natural selection. However, few studies have manipulated progeny environments and demonstrated that the relationship between progeny size and fitness actually differs among environments, and fewer still have demonstrated why selection favors different sized progeny in different environments. We argue that much of the variation in progeny size among females within populations, and among progeny produced by a single female, is probably nonadaptive. However, some species of arthropods exhibit plasticity in progeny size in response to several environmental factors, and much of this plasticity is likely adaptive. We conclude that advances in theory have substantially outpaced empirical data. We hope that this review will stimulate researchers to examine the specific factors that result in variation in selection on progeny size within and among populations, and how this variation in selection influences the evolution of the patterns we observe.