Recent studies with Drosophila have suggested that there is extensive genetic variability for phenotypic plasticity of body size versus food level. If true, we expect that the outcome of evolution at very different food levels should yield genotypes whose adult size show different patterns of phenotypic plasticity. We have tested this prediction with six independent populations of Drosophila melanogaster kept at extreme densities for 125 generations. We found that the phenotypic plasticity of body size versus food level is not affected by selection or the presence of competitors of a different genotype. However, we document increasing among population variation in phenotypic plasticity due to random genetic drift. Several reasons are explored to explain these results including the possibility that the use of highly inbred lines to make inferences about the evolution of genetically variable populations may be misleading.