Growth responses of different tree species to elevated CO 2 vary enormously, even when grown under the same basic experimental conditions (such as nutrient supply and light conditions). A test of the hypothesis that this variation is associated with the species' shade-tolerance is presented. A meta-analysis of 74<space>logarithm-transformed response ratios RR (total biomass in ~700<space>μmol.mol - 1 CO 2 /biomass in current ambient CO 2 ), grouped according to the species' relative tolerance to shade, revealed that RR differed widely and significantly between seedlings or saplings of woody species of different shade-tolerance. Only comparative studies were included in the meta-analysis, and interference from differences in basic experimental conditions between experiments was minimised. The mean RR of species with very high shade-tolerance was 1.51. This was significantly greater than the value of 1.18 for species intolerant of shade. Smaller differences in mean RR were found between groups of species that differed less widely in shade-tolerance. When responses were categorised according to two aspects of light environment in the experiments, daily integrated or peak photosynthetically active photon flux density (PFD), the differences in RR between different shade-tolerance classes disappeared at daily integrated PFD below ~20<space>mol.m - 2 .d - 1 or peak PFD below ~250<space>μmol.m - 2 .s - 1 . Growth stimulation in juvenile trees is likely to cause increased survivorship in subsequent years. Even if the observed differences in RR were restricted to sites experiencing no or only moderate shading, it appears likely that they will affect species composition of naturally regenerating forests in the long term.