Mutations in presenilins 1 and 2 (PS1 and PS2) are responsible for approximately 40% of all early onset familial Alzheimer's disease (FAD) monogenic cases. Presenilins (PSs) function as the catalytic subunit of γ-secretase and support cleavage of the amyloid-β protein precursor (AβPP). We previously discovered that PSs also function as passive endoplasmic reticulum (ER) calcium (Ca2+) leak channels and that most FAD mutations in PSs affected their ER Ca2+ leak function. To further validate the relevance of our findings to human disease, we here performed Ca2+ imaging experiments with lymphoblasts established from FAD patients. We discovered that most FAD mutations in PSs disrupted ER Ca2+ leak function and resulted in increased ER Ca2+ pool in human lymphoblasts. However, we found that a subset of PS1 FAD mutants supported ER Ca2+ leak activity, as ER Ca2+ pool was unaffected in lymphoblasts. Most of the "functional" mutations for ER Ca2+ leak were clustered in the exon 8-9 area of PSEN1 gene and segregated with the cotton wool plaques and spastic paraparesis clinical phenotype occasionally observed in PS1 FAD patients. Our findings with the "functional" and "non-functional" PS1 FAD mutants were confirmed in Ca2+ rescue experiments with PS double-knockout mouse embryonic fibroblasts. Based on the combined effects of the PS1 FAD mutations on ER Ca2+ leak and γ-secretase activities we propose a model that explains the heterogeneity observed in FAD. The proposed model has implications for understanding the pathogenesis of both familial and sporadic AD.