Pharmacological and electrophysiological evidence has shown that opioid receptors are involved in the mechanism of heroin dependence. Thus, opioid receptors are appropriate candidate genes for case‐control association studies of heroin dependence. Previously, two single nucleotide polymorphisms (SNPs), OPRD1 921T > C and 80G > T, of the human δ opioid receptor gene were used in population‐based studies of heroin dependence. One study in a German population found that OPRD1 921T > C was associated with heroin dependence. This finding, however, was not replicated in a different German sample. To test the hypothesis that OPRD1 or a closely linked gene is associated with heroin dependence, we used 5′ nuclease assays to genotype both OPRD1 SNPs in 450 Chinese heroin dependent patients and 304 unaffected controls from the same population. In addition, five SNPs distributed in four other genes: ADH2, ALDH2, OPRM1, and DRD1, were used as genomic control loci to test the case and control populations for stratification bias. Genotype and allele frequencies at OPRD1 921T > C were not significantly different, and the OPRD1 80G was absent from both Chinese opioid dependence patients and controls. Based on the genotype and allele frequencies of the genomic control loci, there was no evidence for stratification bias capable of masking an association of OPRD1 to heroin dependence in this large and homogenous Chinese sample. Therefore, these data do not support an association between the OPRD1 gene and heroin dependence in the Chinese population. Published 2002 Wiley‐Liss, Inc.