Until now, observational studies, randomized controlled trials (RCTs), and Mendelian randomization (MR) studies have explored the impact of vitamin D on Alzheimer's disease (AD), and reported inconsistent findings. In MR studies, the sensitivity analysis by removing GC rs2282679 variant highlighted no association of 25OHD levels with AD risk, which indicates that vitamin D-binding protein (DBP) encoded by GC may have distinct effects on AD risk. Here, we aim to clarify this assumption. We selected the GC rs2282679 variant associated with DBP levels (p = 3.30E-76) as the instrumental variable, and extracted the summary statistics of rs2282679 variant in multipleADGWAS datasets from IGAP, Complex Trait Genetics (CTG) lab, and UK Biobank. We then performed a MR study to investigate the causal association between DBP levels and AD. In IGAP, MR analysis showed that the genetically DBP levels (per 1 standard deviation (SD) increase 50 mg/L) were significantly associated with reduced AD risk (OR = 0.63, 95% CI: 0.45-0.89, p = 0.009). Importantly, the estimates from two sensitivity analyses were consistent with the main estimate in terms of direction and magnitude. Meanwhile, we found no causal association between DBP levels and other four AD phenotypes in CTG lab and UK Biobank. In summary, we highlight the role of DBP levels in AD risk, and provide strong support evidence that DBP may be the therapeutic agent for the treatment of AD. Meanwhile, our findings clarify the assumption that DBP may drive the observed relationship between 25OHD levels and AD.