OBJECTIVE An allelic association between the TaqI "A" system A1 allele at the D2 dopamine receptor locus (DRD2) and either alcoholism or severe alcoholism has been proposed. Our purpose was to evaluate whether, based on all of the accumulated evidence, this association could be considered to be proven. METHODS We considered data from all published reports of DRD2 allele frequency in alcoholics, controls, or both. METHODS We concentrated on the issue of replication. We therefore considered all data reported (on white samples, because DRD2 allele frequency varies by race and ethnicity) since the first report by Blum et al in 1990. RESULTS We analyzed the set of data for differences in allele frequencies between alcoholics and controls, and for heterogeneity among samples. We also investigated the influence of the data from the first group to report an association (including a subsequent report from that group) on the findings. Our analysis shows that, when all studies subsequent to the original study are considered, there is no significant difference in DRD2 A1 allele frequency between alcoholics and controls, there is significant heterogeneity among reported alcoholics and reported controls, and there is no significant difference in DRD2 A1 allele frequency between severe and not severe alcoholics. Also, the two reports of Blum et al account for all of the (nonsignificant) differences seen between controls, alcoholics, and severe alcoholics. CONCLUSIONS In general, heterogeneity among studies (for alcoholics or controls) is considerably greater than differences between alcoholics and controls overall. The findings to date can best be explained by more conservative interpretations than a confirmed physiologically important allelic association between DRD2 alleles and alcoholism. These other possibilities include sampling error and ethnic variation in those studies that individually showed a large effect.