A highly significant population genetic association was reported between alcoholism and a TaqI DNA polymorphism recognized by the dopamine D2 receptor gene (DRD2) probe. We tested this linkage using DRD2 markers in well-characterized U.S. Caucasian alcoholics versus population controls, in Finnish alcoholics and ethically matched nonalcoholics and in Cheyenne Indian alcoholics and controls. No association between the dopamine D2 receptor polymorphism and alcoholism was observed in any of these populations. In U.S. and Finnish Caucasians there was no relationship between D2 marker status and an indicator of central dopaminergic function: CSF homovanillic acid. Linkage disequilibrium between the Taq1 locus and an SSCP variant in the immediate 3' region of the gene was shown to vary between populations and to be only approximately 35% in the two Caucasian populations. thus the population association method may have limited strength to exclude or include this locus. American Indians had a Taq1 allele frequency three times that of Caucasians and Blacks a frequency twice as high, providing a mechanism for spurious associations through inadequate matching of patients and controls.