It was suggested that the gene encoding for sorLa, (SORL1) may affect Alzheimer's disease (LOAD) through a female-specific mechanism. The aims of this study were to confirm the role of gender in modulating the association between SORL1 and LOAD and to ascertain the influence of SORL1 on cognitive impairment, neuropsychiatric symptoms (BPSD) and secretion of pro-inflammatory cytokines.Ninety six outpatients with LOAD and 120 unrelated controls were genotyped for APOE and three SNPs at the 5' end of SORL1(intron 6): SNP 8 (rs668387); SNP 9 (rs68902); SNP 10 (rs641120). Clinical evaluation was made with the MMSE, Neuropsychiatric Inventory (NPI) and Cornell Scale for Depression in Dementia (CDDS). ELISPOT assays were used to measure pro-inflammatory cytokine (TNF-alpha; IL-6; IL-1beta; IFN-gamma) production in peripheral blood mononuclear cell (PBMC) supernatant from AD patients. SORL1 SNPs were not associated with LOAD in overall sample. Instead the G-alleles at SNPs 9 (p=0.015) and 10 (p=0.015) and the CGG haplotype (p=0.02) were associated with LOAD in the women subgroup. The TAA haplotype was marginally protective in AD patients being associated with lower BPSD scores (p=0.01). The same haplotype was also associated with higher IL-1beta (p=0.01) production. These genetic effects were not modified by APOE4 allele and controlled for illness duration and treatment. In conclusion, SORL1 does not appear to be a major risk factor for LOAD. Its contribution could be underestimated in our small sample. Sex-specific factors could modulate the association between SORL1 and AD. The influence of SORL1 variants on production of inflammatory cytokines warrants further investigation.