Costs of phenotypic adaptation to changing environments have often been studied in morphological structures. Such structures typically are irreversible for at least some stage in the organism's life. In this study we investigated whether recurrent and reversible adaptation to changes in the thermal environment incurs a cost in terms of some key life-history traits in the collembolan Orchesella cincta. We exposed juvenile O. cincta to two treatments differing in the frequency of temperature fluctuation but with equal total temperature sums. In the high frequency treatment temperature fluctuated daily between 10 and 20 °C, while in the low frequency treatment temperature fluctuated on a weekly schedule. During the treatments we measured juvenile growth rate and juvenile mortality, and after six weeks the animals were transferred to constant 15 °C and adult starvation resistance was assessed. We found no significant differences between the treatments in juvenile growth rate or juvenile survival. Also, adults that had grown up under high frequency temperature fluctuations did not suffer from reduced starvation resistance compared to animals growing under low frequency temperature fluctuations. This finding supports the hypothesis that selection minimizes the production costs of inducible phenotypes and suggests that the development of optimal phenotypes and evolution of temperature reaction norms are not constrained by such costs. © 2011 Elsevier Ltd.