focusing on G E interactions, logistic regression analyses were performed to evaluate main effects of SNP genotype coded either as number of risk-associated alleles or as separate variables for the presence of one and two alleles. No significant main effects were found for SNP genotype coded in either form, including when CTFS was added to these models. The results of logistic regression analyses that included G E interactions involving CTFS and SNP genotype while controlling for main effects of CTFS and SNP genotype are displayed in Table 1. Findings across SNPs provide consistent support for G E interactions involving CTFS and SNP genotype (those for rs279858 fell just below significance) coded as number of risk-associated alleles. When separate variables were coded for the presence of one or two risk-associated alleles, significant G E interactions are only found for homozygous individuals. Further analyses controlling for DSM-IV nicotine and alcohol dependence diagnoses produced remarkably similar findings (Supplementary Table 1). To ensure that our results are specific to PTSD rather than major depressive disorder (MDD) with which PTSD has considerable comorbidity, we performed similar series of analyses focusing on MDD and found no evidence for significant G E interactions (results not shown). The strength of our findings and their consistency with prior investigations are extremely encouraging. However, our results must be considered preliminary until they are replicated. Animal and human studies suggest that individuals who experience childhood trauma are likely to undergo epigenetic modifications that alter expression of GABAA receptor subunits affecting ligand affinity at the benzodiazepine receptor binding site. The G E interactions that we observed may have resulted from one or more GABRA2 polymorphisms (the four SNPs are known to be in linkage disequilibrium with r values 0.86–0.97) that contributed to PTSD development in those experiencing childhood trauma exposure, perhaps by accentuating the postevent anxiogenic response and facilitating traumatic memory formation. Limitations that must be considered when interpreting our results include the limited number of individuals for whom both phenotypic data and DNAs (hence SNP genotype results) are available. Although their selection through participation in a linkage study of nicotine dependence may have introduced some bias, findings were unchanged in analyses controlling for nicotine dependence. The degree to which our results will be generalizable to samples not enriched for childhood trauma exposure is unclear. Similarly, as a minority of PTSD in our sample arose from traumatic events whose onset occurred after age 17 (28.3%), we do not have adequate power to determine whether our findings reflect risk for incident PTSD with subsequent trauma or primarily for emergent PTSD with childhood trauma. Additional research will be necessary to address these issues and to determine the polymorphism(s) underlying our findings. Our results suggest polymorphisms of genes for which posttrauma expression changes have been observed may provide fertile ground for future genetic studies of PTSD.