Sexual size dimorphism (SSD) is widespread and variable among animals. According to the differential equilibrium model, SSD in a given species is expected to result if opposing selection forces equilibrate differently in both sexes. Here I review the factors that affect the evolution of SSD specifically as they relate to behavior. Taking the approach that SSD results as an epiphenomenon from separate but related selection on male and female body size, the advantages and disadvantages of large size in terms of the standard components of individual fitness (mating success, fecundity, viability, growth, foraging success) are discussed to help guiding future research on the subject. This includes a discussion of intra-SSDs. The main conclusions are: (1) Evidence for disadvantages of large body size is still sparse and requires more research. In contrast, evidence for sexual or fecundity selection favoring large body size is overwhelming, so these mechanisms do no longer require special attention, but need to be documented nonetheless to acquire a complete picture. (2) Some hypotheses suggesting that small size is favored are not well investigated at all, because they apply only to some species or restricted situations, may be difficult to study, or have simply been disregarded. Evidence for these cryptic hypotheses is best revealed using experiments under multiple environmental (food, temperature, etc.) stresses with particularly well-suited model species. (3) The evolution of SSD ultimately depends on processes generating variation within as well as between the sexes, so studies should always investigate and report effects on both sexes separately, in addition to size-dependent effects within each sex; within sexes the key issue is whether small individuals under, over- or perfectly compensate their general fitness disadvantage. (4) Tests of several hypotheses should be integrated in case studies of well-suited model species to investigate selection on body size comprehensively. For example, all episodes of sexual selection (mate search, competition, pre- and post-copulatory choice) should be addressed in conjunction. Investigations of size-selective and sex-dependent predation should take the viewpoint of the prey rather than the predator to permit integration of effects throughout prey ontogeny generated by various predators with differing preferences. Comparative studies should also test multiple alternative hypotheses at the same time to permit stronger inference. (5) Experimental behavioral studies of sexual and natural selection should provide selection differentials using the available standard methods. This would allow integration with phenomenological studies of selection and facilitate subsequent meta-analyses, which are very valuable in evaluating general patterns. (6) Comparative phylogenetic studies identifying patterns and phenomenological and experimental studies of model species that investigate particular mechanisms should be integrated, so that macro-evolutionary patterns can be linked to micro-evolutionary processes, which is the central paradigm of evolutionary ecology. (7) A major problem is the general difficulty of separating causes generating a particular body size and SSD over evolutionary time and its consequences for behavior and ecology today, i.e. today's researchers cannot completely avoid this 'ghost of SSD evolution past'.
