Amyloid and tau pathological accumulation should be considered for Alzheimer's disease (AD) definition and before subjects' enrollment in disease-modifying trials. Although age, APOEε4, and sex influence cerebrospinal fluid (CSF) biomarker levels, none of these variables are considered by current normality/abnormality cutoffs. Using baseline CSF data from 2 independent cohorts (PharmaCOG/European Alzheimer's Disease Neuroimaging Initiative and Alzheimer's Disease Neuroimaging Initiative), we investigated the effect of age, APOEε4 status, and sex on CSF Aβ42/P-tau distribution and cutoff extraction by applying mixture models with covariates. The Aβ42/P-tau distribution revealed the presence of 3 subgroups (AD-like, intermediate, control-like) and 2 cutoffs. The identification of the intermediate subgroup and of the higher cutoff was APOEε4 dependent in both cohorts. APOE-specific classification (higher cutoff for APOEε4+, lower cutoff for APOEε4-) showed higher diagnostic accuracy in identifying MCI due to AD compared to single Aβ42 and Aβ42/P-tau cutoffs. APOEε4 influences amyloid and tau CSF markers and AD progression in MCI patients supporting i) the use of APOE-specific cutoffs to identify MCI due to AD and ii) the utility of considering APOE genotype for early AD diagnosis.