This paper discusses several factors affecting the development of children prenatally exposed to drugs. In the “first generation” of research in this field a main factor model of disease formed the basis for a belief in the feasibility of detecting the direct pharmacological or teratogenic effects of drug exposure on long-term child development. However, the clustering of confounding variables has constituted a major problem in identifying these effects. In the last few years a “second generation” of research in this field has emerged, and investigators have moved beyond simple main-effect models. The importance of controlling for confounding variables has been underscored. However, prenatal substance exposure is still often studied within a teratology model where the main goal is the search for unique effects of a specific drug or substance. Based on this review it is suggested that an appropriate model for understanding the development of drug-exposed children cannot be based on a main-effect perspective. Rather, such a model must evolve from a contextual perspective, and it is suggested that a transactional model, where both potential risk factors and protective factors are considered, should replace the traditional teratology model in this field.