The pro-inflammatory cytokine interleukin(IL)-1beta is a main component in inflammatory pathways and is overexpressed in the brain of Alzheimer's disease (AD) patients. Several studies report associations between IL-1beta polymorphisms and AD, but findings from different studies are controversial. Our aim was to verify the correlation between the single nucleotide polymorphisms (SNPs) of the IL-1beta, at sites -511 and +3953, and AD by meta-analysis. Computerized bibliographic searches of PUBMED and AlzGene database (http://www.alzgene.org) were supplemented with manual searches of reference lists. There is evidence for association between IL-1beta +3953 SNP and AD, with an OR=1.60 (95% C.I.: 1.16-2.22; Z=2.83 p=0.005) for TT genotype. No significant difference in genotype distribution of the IL-1beta -511 SNP in AD was obtained, but high between-study heterogeneity was found. To reduce heterogeneity, subgroup analyses were performed using, as stratifying variables, characteristics of the population under study (age, gender, type of AD diagnosis, Mini Mental State Examination of the controls) and characteristics related to the study design (statistical power of individual studies). The frequency of the IL-1beta -511 TT genotype resulted significantly higher than other genotypes only when the Caucasian studies with the highest statistical power were included in the subgroup analysis (OR=1.32; 95% C.I.: 1.03-1.69; p=0.03), with no evidence of between-study heterogeneity. Our data support an association between the TT genotype of IL-1beta +3953 SNP and AD, and suggest a possible association of the -511 TT genotype. Unreplicability of the results seems to be due mainly to the lack of statistical power of the individual studies.