Setd2 is the only enzyme that catalyzes histone H3 lysine 36 trimethylation (H3K36me3) on virtually all actively transcribed protein-coding genes, and this mechanism is evolutionarily conserved from yeast to human. Despite this widespread and conserved activity, Setd2 and H3K36me3 are dispensable for normal growth of yeast but are absolutely required for mammalian embryogenesis, such as oocyte maturation and embryonic vasculogenesis in mice, raising a question of how the functional requirements of Setd2 in specific developmental stages have emerged through evolution. Here, we explored this issue by studying the essentiality and function of Setd2 in zebrafish. Surprisingly, the setd2-null zebrafish are viable and fertile. They show Mendelian birth ratio and normal embryogenesis without vascular defect as seen in mice; however, they have a small body size phenotype attributed to insufficient energy metabolism and protein synthesis, which is reversable in a nutrition-dependent manner. Unlike the sterile Setd2-null mice, the setd2-null zebrafish can produce functional sperms and oocytes. Nonetheless, related to the requirement of maternal Setd2 for oocyte maturation in mice, the second generation of setd2-null zebrafish that carry no maternal setd2 show decreased survival rate and a developmental delay at maternal-to-zygotic transition. Taken together, these results indicate that, while the phenotypes of the setd2-null zebrafish and mice are apparently different, they are matched in parallel as the underlying mechanisms are evolutionarily conserved. Thus, the differential requirements of Setd2 may reflect distinct viability thresholds that associate with intrinsic and/or extrinsic stresses experienced by the organism through development, and these epigenetic regulatory mechanisms may serve as a reserved source supporting the evolution of life from simplicity to complexity.