Fisher's fundamental theorem states that heritable variation for net fitness sets a limit to the rate of response to natural selection. How will temperate (i.e. cold-tolerant) species cope with contemporary rapid global warming? Using three-fold replicated lines of Drosophila subobscura that had been allowed to evolve for 4 years (between 32 and 59 generations) at 13 degrees C (cold), 18 degrees C (the supposed optimum temperature), and 22 degrees C (warm) I assess here how net fitness changes according to thermal environments. Net fitness was estimated following the classical approach in population genetics of competing over a number of generation in outbred experimental populations multiple wild-type O chromosomes (homologous to arm 3R in D. melanogaster) independently derived from each base thermal stock in an otherwise homogeneous genetic background against a balancer chromosome. Warm-adapted populations ('warm-adapted O chromosomes') performed comparatively well at all tested temperatures. However, net fitness was severely reduced in cold-adapted populations when transferred to warmer conditions. It seems, therefore, that thermal fitness breath for D. subobscura flies is positively associated to temperature. These findings are discussed in relation to the fast world-wide clinal shifts in the frequency of genetic markers correlated with current climate change.