Background: Serum glial fibrillary acidic protein (GFAP) is a biomarker for astrocytic injury and astrogliosis. Concentrations are elevated in numerous neurological disorders, including a pronounced increase in Alzheimer's disease (AD). However, GFAP levels in the serum also increase with age. Consequently, the integration of GFAP levels into clinical routine and their interpretation demands age-adjusted reference values. Methods: Serum from 1273 subjects (952 non-inflammatory and non-neurodegenerative neurological controls and 321 subjects with AD) were analyzed for GFAP using the microfluidic Ella system. Age-dependent serum GFAP reference values were calculated by additive quantile regression analysis. Percentiles and z-scores were employed for the presentation of GFAP levels as a function of age. Results: All patients within the AD continuum exhibited statistically elevated serum GFAP levels in comparison to the control cohort (p<0.0001). This remained the case when the newly generated age-corrected z-scores were applied (p<0.0001). In the control cohort, a non-linear elevation of serum GFAP with increasing age was observed (Spearman correlation coefficient (r) 0.62, 95%CI 0.58-0.66, p<0.0001). In contrast, the AD cohort exhibited a more linear increase in serum GFAP levels (0.16, 95%CI 0.05-0.26, p=0.004). Age-dependent cut-offs for serum GFAP were calculated for different AD age groups. The calculated areas under the curve (AUC 0.97) demonstrated excellent diagnostic test performance in the early onset age group. This effect was less marked in the elderly subjects (AUC 0.72). Conclusions: Our novel GFAP z-scores enable the integration and interpretation of serum GFAP levels in clinical practice, moving from group to individual level. They support both intra- and interindividual interpretation of single GFAP levels in neurological diseases with astrocytic pathology, including an accurate discrimination of Alzheimer's disease.