Calcium signaling in the brain is fundamental to the learning and memory process and there is evidence to suggest that its dysfunction is involved in the pathological pathways underlying Alzheimer's disease (AD). Recently, the calcium hypothesis of AD has received support with the identification of the non-selective Ca(2+)-permeable channel CALHM1. A genetic polymorphism (p. P86L) in CALHM1 reduces plasma membrane Ca(2+) permeability and is associated with an earlier age-at-onset of AD. To investigate the role of CALHM1 variants in early-onset AD (EOAD), we sequenced all CALHM1 coding regions in three independent series comprising 284 EOAD patients and 326 controls. Two missense mutations in patients (p.G330D and p.R154H) and one (p.A213T) in a control individual were identified. Calcium imaging analyses revealed that while the mutation found in a control (p.A213T) behaved as wild-type CALHM1 (CALHM1-WT), a complete abolishment of the Ca(2+) influx was associated with the mutations found in EOAD patients (p.G330D and p.R154H). Notably, the previously reported p. P86L mutation was associated with an intermediate Ca(2+) influx between the CALHM1-WT and the p.G330D and p.R154H mutations. Since neither expression of wild-type nor mutant CALHM1 affected amyloid ß-peptide (Aß) production or Aß-mediated cellular toxicity, we conclude that rare genetic variants in CALHM1 lead to Ca(2+) dysregulation and may contribute to the risk of EOAD through a mechanism independent from the classical Aß cascade.