Background. An adoption study of alcoholism suggests that in women, the impact of genetic risk factors become greater in the presence of conflict in the family of origin. Is the same true for cigarette smoking (CS)?

Method. We obtained, in a sample of 1676 twins from female–female twin pairs from a population-based register, a measure of maximum lifetime CS (divided into six ordinal categories) and family dysfunction (FD) assessed as the mean report of up to four informants (twin, co-twin, mother, father). Statistical analysis was conducted by traditional regression analysis and a moderator structural equation twin model using the computer program Mx.

Results. With increasing levels of FD, maximum CS increased substantially while correlations for CS in monozygotic (MZ) and dizygotic (DZ) twins decreased modestly. Regression analyses demonstrated reduced twin-pair resemblance for CS with increasing levels of FD. The best-fit structural equation model found high levels of heritability for CS and no evidence for a role of shared environment. With increasing levels of FD, the proportion of variance in CS due to genetic factors (i.e. heritability) decreased while that due to unique environmental effects increased.

Conclusions. Several different statistical methods suggested that, contrary to prediction, heritability of CS decreased rather than increased with higher levels of dysfunction in the family of origin. The hypothesis that genetic effects for psychiatric and drug-use disorders become stronger in more adverse environments is not universally true.