Understanding the mechanisms that determine an individual's sex remains a primary challenge for evolutionary biology. Chromosome-based systems (genotypic sex determination) that generate roughly equal numbers of sons and daughters accord with theory, but the adaptive significance of environmental sex determination (that is, when embryonic environmental conditions determine offspring sex, ESD) is a major unsolved problem. Theoretical models predict that selection should favour ESD over genotypic sex determination when the developmental environment differentially influences male versus female fitness (that is, the Charnov-Bull model), but empirical evidence for this hypothesis remains elusive in amniote vertebrates--the clade in which ESD is most prevalent. Here we provide the first substantial empirical support for this model by showing that incubation temperatures influence reproductive success of males differently than that of females in a short-lived lizard (Amphibolurus muricatus, Agamidae) with temperature-dependent sex determination. We incubated eggs at a variety of temperatures, and de-confounded sex and incubation temperature by using hormonal manipulations to embryos. We then raised lizards in field enclosures and quantified their lifetime reproductive success. Incubation temperature affected reproductive success differently in males versus females in exactly the way predicted by theory: the fitness of each sex was maximized by the incubation temperature that produces that sex. Our results provide unequivocal empirical support for the Charnov-Bull model for the adaptive significance of temperature-dependent sex determination in amniote vertebrates.