The primitive land plant life cycle featured the production of spores of unimodal size, a condition called homospory. The evolution of bimodal size distributions with small male spores and large female spores, known as heterospory, was an innovation that occurred repeatedly in the history of land plants. The importance of desiccation-resistant spores for colonization of the land is well known, but the adaptive value of heterospory has never been well established. It was an addition to a sexual life cycle that already involved male and female gametes. Its role as a precursor to the evolution of seeds has received much attention, but this is an evolutionary consequence of heterospory that cannot explain the transition from homospory to heterospory (and the lack of evolutionary reversal from heterospory to homospory). Enforced outcrossing of gametophytes has often been mentioned in connection to heterospory, but we review the shortcomings of this argument as an explanation of the selective advantage of heterospory. Few alternative arguments concerning the selective forces favouring heterospory have been proposed, a paucity of attention that is surprising given the importance of this innovation in land plant evolution. In this review we highlight two ideas that may lead us to a better understanding of why heterospory evolved. First, models of optimal resource allocation - an approach that has been used for decades in evolutionary ecology to help understand parental investment and other life-history patterns - suggest that an evolutionary increase in spore size could reach a threshold at which small spores yielding small, sperm-producing gametophytes would return greater fitness per unit of resource investment than would large spores and bisexual gametophytes. With the advent of such microspores, megaspores would evolve under frequency-dependent selection. This argument can account for the appearance of heterospory in the Devonian, when increasingly tall and complex vegetative communities presented competitive conditions that made large spore size advantageous. Second, heterospory is analogous in many ways to anisogamy. Indeed, heterospory is a kind of re-invention of anisogamy within the context of a sporophyte-dominant land plant life cycle. The evolution of anisogamy has been the subject of important theoretical and empirical investigation. Recent work in this area suggests that mate-encounter dynamics set up selective forces that can drive the evolution of anisogamy. We suggest that similar dispersal and mating dynamics could have underlain spore size differentiation. The two approaches offer predictions that are consistent with currently available data but could be tested far more thoroughly. We hope to re-establish attention on this neglected aspect of plant evolutionary biology and suggest some paths for empirical investigation.