To analyze the CYP2D6 allele frequencies in a Japanese population and to evaluate the effects of CYP2D6 variants on in vivo CYP2D6 activity as measured by the dextromethorphan metabolic ratio (MR). Ninety-eight unrelated, healthy Japanese men were phenotyped with dextromethorphan and genotyped by the polymerase chain reaction amplification method for 7 CYP2D6 alleles. The CYP2D6*1, CYP2D6*10, CYP2D6*2, CYP2D6*5, CYP2D6*4, and CYP2D6*21 allele frequencies in our Japanese subjects were 0.423, 0.408, 0.092, 0.061, 0.020, and 0.010, respectively. Thirty-three subjects (33.7%) were heterozygous for *10/*1, and 18 (18.4%) and 17 (17.3%) subjects were homozygous for *1 and *10, respectively. Subjects who were homozygous for *10 showed the highest dextromethorphan MR among these 3 genotypes. Eighteen subjects (18.3%) were heterozygous for *2, but their dextromethorphan MR values were not greater than the MR values of subjects who were homozygous for *1. One subject was a poor metabolizer phenotypically, and he was homozygous for *5. The CYP2D6 allele frequencies in our Japanese subjects differed from those determined in previous studies of white subjects or mainland Chinese subjects. Individuals homozygous for *10 who have relatively low in vivo CYP2D6 activity represent almost 20% of the Japanese population. In addition, we did not identify any subjects with amplified *2 among our 98 Japanese men.