In ectotherms there is typically a strong and positive correlation between growth rate and ambient temperature when food is not limiting. However, the exact relationship between growth rate and temperature varies among populations in many species. As a consequence, it has been suggested that selection for a rapid increase in growth rate with temperature should be stronger in populations experiencing a high degree of time–stress, compared with populations experiencing little time–stress. In the present study we take this adaptive hypothesis further and investigate if variation in time–stress among individuals of a single population may affect the relationship between growth rate and ambient temperature. Time–stress was manipulated by rearing larvae of the butterfly Lasiommata maera in different day–length regimes. The results show that individuals experiencing a higher degree of time–stress increase their growth rates more in higher temperatures compared with individuals under less time–stress. Hence, the adaptive hypothesis was supported and the relationship between growth rate and temperature was highly state dependent. These findings may be of general importance for understanding the evolution of life histories in seasonal environments.