A major concern with the vast literature associating the highly polymorphic 48 bp VNTR in exon III of the human dopamine receptor D4 gene (DRD4) with various behavioral phenotypes is the lack of concordance between studies. Part of the problem arises from the absence of a universally accepted scheme for pooling the large number of low frequency genotypes into appropriate categories. Here, we investigated the effect of different pooling strategies and genetic models on the reported association between DRD4-exIII-VNTR polymorphism and cigarette smoking. Genotyping was performed on a large randomly selected community-based sample of 2,274 individuals aged 20–24 years. Participants were grouped into sub-samples based on their genotypes to test specific genetic models. Multiple regression analyses were used to assess the relationship between DRD4-exIII-VNTR genotype and cigarette smoking measures while controlling for confounders. While smoking status and age at start of smoking were not associated with the genotype, a significantly (P = 0.006) higher rate of cigarette consumption was observed among carriers of the 7-repeat (7r) allele. Thus, 7r carriers were not more likely to be smokers but if they did smoke they consumed significantly more cigarettes per day than 4r carriers. Unlike previous studies this association was observed only when comparing carriers of the 7r with the 4r but not the other repeat alleles. Our study demonstrates the need for caution when grouping functionally different DRD4-exIII-VNTR alleles in association studies. It particularly highlights the requirement for better functional characterization of the DRD4-exIII-VNTR alleles for interpreting results from association studies. © 2011 Wiley-Liss, Inc.