Cytochrome P450 2B6 (CYP2B6) metabolizes a number of therapeutic drugs and its metabolic activity varies markedly in human liver. Although genetic polymorphisms of CYP2B6 have been reported in noncoding and coding regions, little information is available regarding single nucleotide polymorphisms (SNPs) and their haplotypes in noncoding regions in Asians. Fourteen previously reported SNPs were determined by polymerase chain reaction-restriction fragment length polymorphism or SNaPshot analysis in a Korean population and their haplotypes were inferred from genotype data using an expectation-maximization algorithm. The most common haplotypes were haplotype I, the reference sequence (frequency 0.35), haplotype II (0.19), haplotype III (0.19), and haplotype V (0.12), which together accounted for 85% of all haplotypes. The frequency of haplotype III, which contains -2320C, -1778G, -1186G, -750C, and 15582T, was found to be 2.4-fold higher than that of the *1J allele in Caucasians, and the frequency of haplotype V, which contains -8207C, -1456C, -750C, 516T, and 785G, was 55% of that of the *6B allele in Caucasians. Moreover, haplotype V, the *6B allele, appeared to be completely linked to -8207 within a putative nuclear receptor binding motif, suggesting that lower expressions of the *6B allele may be associated with the presence of noncoding SNPs such as -8207G>C linked to nonsynonymous SNPs. In conclusion, we found 11 previously described polymorphisms and identified four major haplotypes of CYP2B6 in Koreans. The frequencies of the *1J or *6B alleles, which may reduce CYP2B6 enzyme expression, were found to be significantly different between Koreans and Caucasians.