CYP2D6 genotyping (CYP2D6*3, CYP2D6*4, CYP2D6*5, CYP2D6*13, CYP2D6*16 alleles and gene duplications) was previously performed on 1053 Caucasian and African-American lung cancer cases and control individuals and no significant difference in allele frequencies between cases and control individuals detected. We have carried out additional genotyping (CYP2D6*6, CYP2D6*7, CYP2D6*8, CYP2D6*9, CYP2D6*10, CYP2D6*17 alleles) and debrisoquine phenotyping on subgroups from this study to assess phenotype-genotype relationships. African-Americans showed significant differences from Caucasians with respect to frequency of defective CYP2D6 alleles, particularly CYP2D6*4 and CYP2D6*5. The CYP2D6*17 allele occurred at a frequency of 0.26 among 87 African-Americans and appeared to explain higher average metabolic ratios among African-Americans compared with Caucasians. CYP2D6*6, CYP2D6*8, CYP2D6*9 and CYP2D6*10 were rare in both ethnic groups but explained approximately 40% of higher than expected metabolic ratios among extensive metabolizers. Among individuals phenotyped with debrisoquine, 32 out of 359 were in the poor metabolizer range with 24 of these (75%) also showing two defective CYP2D6 alleles. Additional single strand conformational polymorphism analysis screening of samples showing large phenotype-genotype discrepancies resulted in the detection of three novel polymorphisms. If subjects taking potentially interfering drugs were excluded, this additional screening enabled the positive identification of 88% of phenotypic poor metabolizers by genotyping. This sensitivity was comparable with that of phenotyping, which identified 90% of those with two defective alleles as poor metabolizers.