When females mate multiply, male reproductive success depends not only on mating success but also on fertilization success, which is mediated by postcopulatory processes like sperm competition and cryptic female choice. Although postcopulatory sexual selection has the potential to be a major force in driving evolution, very few studies have estimated its strength in the wild, or measured it in such a way as to enable a quantitative comparison with precopulatory sexual selection. Likewise, though polyandry is widespread across taxa and is the focus of a growing body of research, estimates of natural female mating rates are still limited in number. I used extensive behavioral observations of a semi-natural population of Hawaiian swordtail crickets, Laupala cerasina, combined with molecular paternity assignment, to quantify pre- and postcopulatory selection in this population (Chapter 1). The opportunity for postcopulatory selection was over four times as great as for precopulatory selection. To corroborate the patterns I found in this experiment, I also genotyped the sperm stores and offspring of a group of wild adult females, estimating the number of males each female mated with, the number of males that sired her offspring, and the paternity skew among these sires (Chapter 2). Both of these studies revealed that postcopulatory selection is strong in this population, supporting the hypothesis that such selection is a major component of overall sexual selection. Chapter 3 concerns strategic sperm allocation, a form of postcopulatory mate choice whereby males differentially allocate their sperm based on female quality or mating status in order to maximize fitness. Many theoretical models of sperm allocation make assumptions that limit their applicability, such as the common assumption that females mate only twice. Furthermore, many empirical tests of these models fail to make a priori predictions, since the species-specific values of the model parameters that dictate these predictions (e.g., female mating rate and sperm precedence pattern) are often unmeasured. I designed a broadly applicable model that is appropriate for multiply, sequentially mating animals. The model's predictions were met in L. cerasina and in the two other species for which all the relevant data have been published.