Many organisms can adjust their phenotypes to match local environmental conditions via shifts in developmental trajectories, rather than relying on changes in gene frequencies wrought by natural selection. Adaptive developmental plasticity confers obvious benefits in terms of rapid response and higher mean fitness, so why is it not more common? Plausibly, adaptive plasticity also confers a cost; reshaping the phenotype takes time and energy, so that canalised control of trait values enhances fitness if the optimal phenotype remains the same from one generation to the next. Although this idea is central to interpreting the fitness consequences of adaptive plasticity, empirical data on costs of plasticity are scarce. In Australian tiger snakes, larger relative head size enhances maximal ingestible prey size on islands containing large prey. The trait arises via adaptive plasticity in snake populations on newly colonised islands but becomes genetically canalised on islands where snakes have been present for much longer periods. We experimentally manipulated relative head size in captive neonatal snakes to quantify the costs of adaptive plasticity. Although small-headed snakes were able to increase their head sizes when offered large prey, the delay in doing so, and their inability to consume large prey at the outset, significantly reduced their growth rates relative to conspecifics with larger heads at the beginning of the experiment. This study describes a proximate cause to the post-colonisation erosion of developmental plasticity recorded in tiger snake populations.