Background:  It has been proposed that some of the variability in reporting of associations between attention deficit hyperactivity disorder (ADHD) and candidate genes may result from mixing of genetically heterogeneous forms of ADHD using DSM‐IV criteria. The goal of the current study is to test whether population‐based ADHD subtypes defined by latent class analysis help resolve issues of variable findings across individual gene association studies.Methods:  Three studies which had previously reported no associations between polymorphisms of the DRD4 and DAT genes and DSM‐IV defined ADHD were reanalyzed using population‐based and DSM‐IV defined ADHD subtypes.Results:  Across studies no significant associations were found for either DRD4 or DAT polymorphisms using DSM‐IV ADHD subtypes. In contrast, a significant association was found between the combined data set for the 440 base pair 3′ DAT VNTR polymorphism and population‐defined severe combined ADHD (OR = 1.25, p = .01). A marginally significant association was also found between the 7 repeat DRD4 allele and population‐defined severe combined ADHD.Conclusion:  Use of alternative population‐based defined ADHD subtypes may help resolve some of the variable results presented for candidate gene association studies in ADHD.