Background. Despite the demonstrable influence of both genes and the family environment on children's behavioural and emotional development, the mechanisms by which these factors are transmitted from parents to their children are not known. Numerous aspects of the family have long been associated with behavioural and emotional problems in children; it is not clear, however, whether these family variables represent genuine environmental risks or secondary consequences of the underlying genetic liability shared between parents and their children.

Method. In this study we present a model for analysing the non-genetic contributions of family background to risk for childhood and adolescent depression and conduct disturbance using simulated data on adult MZ and DZ twins, their spouses and children.

Results. The twin offspring design provides substantial power to detect remarkably small non-genetic effects on parent–offspring resemblance against the background of genetic transmission. As presented, the model is able to resolve the direction of transmission from both parent to child (passive genotype–environment correlation) and child to parent (evocative genotype–environment correlation).

Conclusions. Unlike many other genetic studies, a study of twins and their children can sort out which putative family environmental risk factors do actually have a significant environmental impact on the child and which ones only appear to do so because they are associated with genetic mediation.