Body size explains most of the variation in fitness within animal populations and is therefore under constant selection from ecological and reproductive pressures, which often promote its evolution in sex-specific directions, leading to sexual size dimorphism (SSD). Several hypotheses have been proposed to explain the vast diversity of SSD across species. These hypotheses emphasize (a) the mate competition benefits to larger male size (sexual selection); (b) the benefits of larger female size for fecundity (fecundity selection); (c) the simultaneous benefits of niche divergence for males and females to reduce intersexual competition for ecological resources (natural selection); and (d) the underlying impact of geographical variation in climatic pressures expected to shape large-scale patterns of SSD in synergy with the above selection pressures (e.g., intensification of fecundity selection as breeding seasons shorten). Based on a new, global-scale amphibian dataset, we address the shortage of large-scale, integrative tests of these four hypotheses. Using a > 3,500 species dataset spanning body size, ecological, life-history, geographical and climatic data, we performed phylogenetic linear models to address the sexual, fecundity, ecological and climatic hypotheses of SSD. Evolution of SSD is discordant between anurans and salamanders. Anuran SSD is shaped by climate (male-biased SSD increases with temperature seasonality) and by nesting site. In salamanders, SSD converges across species that occupy the same types of microhabitats (“ecodimorphs”), whereas reproductive or climatic pressures have no effects on their SSD. These contrasts are associated with latitudinal gradients of SSD in anurans, but not in salamanders. Amphibian SSD is driven by ecological and climatic pressures, whereas no roles for sexual or fecundity selection were detected. We show that macroevolutionary processes determined by different forms of selection lead to latitudinal patterns of trait diversity, and the lack of them.