The recent "overhead threshold" model for optimal age and body size at maturity (Day and Rowe 2002 ) predicts that phenotypic variability in adult body size will be low under inferior environmental quality and will increase with improving conditions. The model is, however, based on a potentially restrictive assumption of a monotone increase of fecundity with increasing body size. On the basis of a numerical model, we show that introducing the concept of maximum adult body size changes the predictions of the model. The dependence of variability in adult body size on environmental quality becomes a concave function with a maximum at intermediate values. Depending on the range of environmental conditions considered, one may therefore expect to observe both increasing and decreasing functions. We test the predictions of our model on a literature-based database of 131 insect species covering all major orders. We demonstrate that, in most species, relative phenotypic variation in body size decreases when environment-specific average of adult body size increases. In the majority of cases at least, such a relationship can be interpreted as a decreased relative variation in better growing conditions. With some potentially meaningful exceptions (e.g., females of capital-breeding insects), the general pattern was largely invariable across different taxa, ecological subdivisions, and sexes.