Production of more than one morphological type of sperm in a common testis has been documented for a variety of invertebrates, including gastropods, spiders, centipedes, and insects. This unusual phenomenon is difficult to explain by current theory, particularly since available evidence indicates that one sperm type is often incapable of effecting fertilization. In this review we critically examine evidence on the distribution and development of sperm heteromorphisms among insects in light of competing hypotheses for the evolutionary origin, maintenance, and function of a non-fertilizing class of sperm. To date, no single hypothesis, including alternatives which assume non-fertilizing sperm are non-adaptive, or that they provision, facilitate, or compete with fertilizing sperm, has received strong empirical support by any group of insects. The diversity of sperm heteromorphisms suggests that non-fertilizing sperm may have different functions in different clades or even serve multiple functions within a clade. We suggest that insight could be gained from (1) new models for the evolution of sperm polymorphism, (2) comparative studies that focus on multiple traits simultaneously (e.g. sperm number, proportion, length, and remating rate) and utilize clades in which more than one gain or loss of sperm heteromorphism has been documented (e.g. Pentatomidae, Carabidae, or Diopsidae), and (3) experimental studies that exploit individual variation or directly manipulate the composition of the male ejaculate.