Termites, ants and earthworms are considered as soil engineers because of their effects on soil properties and their influence on the availability of resources for other organisms, including microorganisms and plants. However, the links between their impacts on the soil environment and the resulting modification of natural selection pressures on engineer as well as on other organisms have received little attention. We suggest that the strategy evolved by species, or functional groups to control their environment can explain the difference between extended phenotype and accidental engineers. Extended phenotype engineers concentrate their activities on the building of biogenic structures in order to maintain optimal conditions for their growth. Conversely, accidental engineers expend energy in moving through the soil to be as close as possible to their optimal environment.We discuss in this paper why termites and ants are best viewed as extended phenotype engineers, and particularly how their impacts on ecosystem functioning can be considered as results of their requirements and how their activities could induce feedback loops affecting themselves. We also focus on the engineering activities of earthworms and discuss why some species can be considered as extended phenotype while others are more accidental engineers. Finally, we discuss links between the strategies developed by engineers and ecosystem functioning. Although both types of engineer create ‘‘hot-spots'' in soil, we argue that extended phenotype engineers have a greater effect on the maintenance of ecosystem heterogeneity since they concentrate their activities at a few points, as compared to accidental engineers which may move through the soil and thus contribute to homogenisation of nutrient distribution throughout the whole ecosystem.