Abstract Synaptic loss is an early hallmark of many neurological disorders, including Alzheimer’s disease, but also occurs in aging brains as evidenced by PET- and CSF-based biomarker studies. This cross-sectional study investigates how blood-based synaptic proteins and other biomarkers relate to synaptic density in brains of older adults without dementia, and how these associations are mediated by gray matter (GM) loss. Plasma levels of synaptic biomarkers including synaptosomal-associated protein of 25 kDa and vesicle-associated membrane protein 2 as well as amyloid, tau, neurodegeneration and neuroinflammation (ATN(I)) biomarkers including the amyloid-β1-42/amyloid-β1-40 ratio, phosphorylated tau181, neurofilament light and glial fibrillary acidic protein were quantified in 61 older adults without dementia [mean age ± standard deviation = 71 ± 6 years, median Mini-Mental State Examination score (interquartile range) = 29 (3), 64% female, 38% late-life depression] who underwent synaptic vesicle glycoprotein 2A PET and T1-weighted MRI. Subsets underwent amyloid (n = 49) and tau (n = 52) PET. The study population demonstrated limited PET-based amyloid-β and tau pathology, which did not associate with any of the investigated plasma biomarkers. Synaptic plasma biomarkers correlated with each other [Spearman’s ρ (95% confidence interval) 0.37 (0.14–0.57), P = 0.019], but not with the ATN(I) plasma biomarkers. Plasma vesicle-associated membrane protein 2 associated with synaptic vesicle glycoprotein 2A PET within frontal, temporal, and occipital cortices [βs (95% confidence interval) −0.75 (−1.06 to −0.44)], independent of age. Glial fibrillary acidic protein also demonstrated associations with synaptic vesicle glycoprotein 2A PET, particularly within temporal regions, the cingulate gyrus, and the precuneus [βs (95% confidence interval) −0.30 (−0.48 to −0.12)] when corrected for age and additionally within the caudate nucleus and thalamus when not [βs (95% confidence interval) −0.45 (−0.64 to −0.25)]. Similar associations were found in subgroups without, respectively, Alzheimer’s disease pathology, or late-life depression (with cognitive impairment). No associations with synaptic vesicle glycoprotein 2A PET were found for synaptosomal-associated protein of 25 kDa, phosphorylated tau181, neurofilament light or amyloid-β1-42/amyloid-β1-40, likely due to the minimal presence of Alzheimer’s disease pathology in the study population. GM volume associated with glial fibrillary acidic protein [βs (95% confidence interval) −0.15 (−0.25 to −0.05)] and partially (29%) mediated its association with synaptic vesicle glycoprotein 2A PET. In conclusion, plasma levels of glial fibrillary acidic protein and vesicle-associated membrane protein 2 may reflect synapse pathology independent of age and beyond general neuronal loss, even in absence of detectable Alzheimer’s disease pathology.