Vitamin D has been investigated in association with cognitive function in older adults. It is unclear whether hypovitaminosis D could be associated with Alzheimer's disease (AD). Our objective was to systematically review and quantitatively synthesize the association of low serum 25-hydroxyvitamin D (25OHD) concentrations with AD in adults. A Medline and PsycINFO® search was conducted on May 2012, with no limit of date, using the MeSH terms "Vitamin D" OR "Hydroxycholecalciferols" combined with the MeSH terms "Alzheimer disease" OR "Dementia" OR "Cognition" OR "Cognition disorders" OR "Memory" OR "Memory Disorders" OR "Executive Function" OR "Attention" OR "Neuropsychological Tests". Of the 284 selected studies, 10 observational studies (including 9 case-controls and 1 cohort study) met the selection criteria. All were of good quality. The number of AD cases ranged from 20 to 211 (40%-100% female). Finally, 7 case-control studies were eligible for fixed and random-effects meta-analyses of bias-corrected effect size of the difference in serum 25OHD concentrations between AD cases and controls using an inverse-variance method. The pooled effect size in random-effects meta-analysis was 1.40 (95% CI: 0.26;2.54), a 'large' effect size that indicates that serum 25OHD concentrations were 1.4 standard deviation units lower in AD cases compared to cognitively healthy controls (p = 0.016). In conclusion, AD cases had lower serum vitamin D concentrations than matched controls. This reinforces the conceptualization of vitamin D as a 'neurosteroid hormone' and as a potential biomarker of AD.