We analysed the impact of the TaqI A1 allele of the D2 dopamine receptor gene on the risk for alcoholism, trying to depict three explanations frequently proposed to explain discrepancies in association and linkage studies: that the A1 allele may act as a marker rather than as a vulnerability factor, that stratification biases and unevaluated controls may explain positive results, and that the A1 allele is modifying the phenotype rather than increasing the risk for alcoholism. We thus tested another (dinucleotide STRP) marker within the DRD2 gene, selected a new homogenous sample of 113 alcoholic patients and 49 unaffected controls strictly matched for ethnic origins, and systematically assessed both samples with a semi-structured interview to detect (in both samples) alcohol dependence, but also such related traits as specificities of complications. The frequency of the A1 allele was not significantly different between alcoholics and controls but when comparing different subgroups of alcoholics, the A1 allele was significantly more frequent in alcoholic patients with somatic complications (OR = 3.00, CI[1.37-6.62]), social and professional complications (OR = 2. 72, CI[1.25-5.90]), or with co-morbid dependence (OR = 2.88, 95% IC [1.16-7.15]). The association for co-morbid dependence and somatic complications was also positive when taking into consideration both STRP and TaqIA polymorphisms. The A1 allele does not increase the risk for alcoholism per se in our sample, but may be involved in a related trait which is partially dependent on the diagnosis of alcoholism, through a disequilibrium with another close mutation.