CALHM1 gene coding non-synonymous SNP P86L (rs2986017) was reported to increase the risk of Alzheimer's disease (AD) in a recent study. We have investigated this genetic variant in 2470 individuals from Spain to conduct an independent replication study of the proposed SNP marker. By applying a recessive model, we observed weak evidence of an association between P86L mutation and late-onset AD (LOAD) susceptibility in our case-control study (OR =1.38 C.I. = [1.01-1.89]). Meta-analysis of available studies also supports a recessive model for CALHM1 P86L variant and provides evidence of between study heterogeneity. Importantly, we found that adjusted mean age at AD onset in P86L homozygous LOAD patients was significantly earlier that in the rest of patients (77.01 +/- 6.1 for P86L homozygous carriers versus 79.0 +/- 6.0 for the rest of patients, p=0.002). We concluded that the CALMH1 gene may contribute to AD risk in our study population. The observed genetic model (recessive) and the estimated magnitude of the effect both imply that virtually all studies performed to date were markedly underpowered to detect this effect and underscore the importance of follow up, replication, and meta-analyses of promising genetic signals.