Single nucleotide polymorphisms (susceptibility genetics) and genomic point mutations (mendelian genetics) can be used in Alzheimer's disease (AD) for diagnostic, predictive and therapeutic purposes. Using a matrix genetic model, including APOE, PS1 and PS2 allelic variants, we have studied the distribution of 36 different genotypes in the AD population (N=479) and the genotype‐related cognitive response to a multifactorial therapy in AD patients with mild‐to‐moderate dementia. The 10 most frequent AD genotypes are the following: 1) E33P112P2+ (17.75%), 2) E33P112P2− (15.55%), 3) E33P111P2+ (10.85%), 4) E34P112P2+ (9.60%), 5) E34P112P2− (7.56%), 6) E33P111P2− (7.10%), 7) E34P111P2+ (4.80%), 8) E33P122P2+ (4.38%), 9) E34P111P2− (4.18%), and 10) E34P122P2+ (3.55%). APOE‐4/4‐related genotypes represent less than 3% in the following order: E44P112P2+> E44P111P2+=E44P111P2−>E44P112P2+>E44P122P2+= E44P122P2−. Multifactorial therapy with CDP‐choline (1000 mg/day)+piracetam (2400 mg/day)+anapsos (360 mg/day) did improve mental performance during the first 6–15 months in a genotype‐specific fashion. The best responders in the APOE series were patients with APOE‐3/4 genotype (r=+0.013), while the worst responders were APOE‐4/4 patients (r=−0.93). PS1‐related genotypes responded in a similar manner; and patients with a defective PS2 gene exon 5 (PS2+) always showed a poorer therapeutic response than PS2− patients. All these data suggest that the therapeutic outcome in AD exhibits a genotype‐specific pattern, and that a pharmacogenomic approach to AD might be a valuable strategy for drug development and monitoring.