The relationship between genetic variation in CYP2D6 and variable drug response represents a potentially powerful pharmacogenetic tool. However, little is known regarding this relationship in the genetically diverse South African population. The aim was therefore to evaluate the relationship between predicted and measured CYP2D6 phenotype. An XL-PCR+Sequencing approach was used to determine CYP2D6 genotype in 100 healthy volunteers and phenotype was predicted using activity scores. With dextromethorphan as the probe drug, metabolic ratios served as a surrogate measure of in vivo CYP2D6 activity. Three-hour plasma metabolic ratios of dextrorphan/dextromethorphan were measured simultaneously using semi-automated online solid phase extraction coupled with tandem mass spectrometry. Partial adaptation of the activity score system demonstrated a strong association between genotype and phenotype, as illustrated by a kappa value of 0.792, inter-rater discrepancy of 0.051 and sensitivity of 72.7%. Predicted phenotype frequencies using the modified activity score were 1.3% for poor metabolisers (PM), 7.6% for intermediate metabolisers (IM) and 87.3% for extensive metabolisers (EM). Measured phenotype frequencies were 1.3% for PM, 13.9% for IM and 84.8% for EM. Comprehensive CYP2D6 genotyping reliably predicts CYP2D6 activity in this South African cohort and can be utilised as a valuable pharmacogenetic tool.