Our post hoc analysis examining smokers and never-smokers was limited in its power to determine whether the association with alcohol in this sample is driven primarily by the smokers, because of the reduced sample size of the never-smoking group. Future studies examining a larger sample of non-smokers for association of these polymorphisms and alcohol phenotypes are merited. If association of CHRNA6 and CHRNB3 with alcoholism is found to be limited to only smokers it raises some interesting questions as to whether there is an underlying simultaneous genetic effect on both phenotypes, or whether the association is truly with smoking and other factors lead to a correlation between smoking and alcoholism. It is interesting that in previous associations of nicotine behaviors with CHRNB3/A6, the strongest signals have been found in the putative promoter and 5'UTR of CHRNB3 or in exon 6 of CHRNA6. In this study, the association with alcohol is with SNPs in intron 2 of CHRNA6 and intron 1 of CHRNB3, with only a moderate level of association with exon 6 of CHRNA6 and no association with the upstream
