Insect galls are dramatic examples of extended phenotypes: although composed of host plant tissues, their development is largely controlled by insect genes. Adaptive explanations for gall traits should thus be expressed in terms of impacts on insect fitness, but the extent to which interspecific variation in gall structure is adaptive, and the possible selective pressures driving diversification in gall form remain controversial. In colonial aphids and thrips, gall structures probably diversified in response to selection for enhancement of the surface area available for feeding. In other taxa, such as gall wasps and gall midges, diversity is expressed predominantly in non-nutritive tissues, particularly those on the gall surface. All natural enemies attack the occupants of closed galls by penetrating gall tissue, and modifications that reduce enemy attack rates should thus be favoured. Recent studies of intraspecific variation in gall form strongly support a defensive role for several traits, but, to date, there is little empirical support for enemies as a cause of interspecific variation in gall form. Selection imposed by enemies nevertheless remains the most probable adaptive explanation for the evolution of diversity. We suggest that this hypothesis has yet to be tested explicitly, and discuss the requirements for an appropriate cross-species analysis.