OBJECTIVES: To quantify the potential value of cerebrospinal fluid (CSF) biomarker testing for patients with mild cognitive impairment (MCI). Biomarkers provide information about a patient’s risk of developing AD and can allow for early targeted interventions for those patients found to be at higher risk of AD than others.  METHODS: We developed a state-transition Markov model to project lifetime AD-free life years, costs and quality-adjusted life years (QALYs). We conducted a cost-effectiveness analysis of using CSF biomarker testing combined with the subsequent treatments to delay the clinical diagnosis of AD (test-treat) compared to no treatment, and treatment strategies. For the test-treat strategies, we considered treating by different levels of risk, varying from treating only the highest risk group to treating all but the lowest risk group (total 5 risk levels). We performed deterministic and probabilistic sensitivity analyses (PSA) and conducted an expected value of perfect information (EVPI) analysis to estimate the value of eliminating uncertainty of all parameters. RESULTS: We found that treating MCI patients by their risk levels produced extra 0.9-3.8 AD-free life months compared to no treatment. Three out of four test-treat strategies were ruled out by extended dominance. No treatment resulted in the highest cost and the highest effectiveness, with an incremental cost-effectiveness ratio of $30,000 per QALY compared to treating all patients. No treatment was optimal in 63% of the PSA iterations over 37% of the treatment strategy at willingness to pay of $50,000/QALY. The total EVPI was $3,512 per patient. CONCLUSIONS: This study illustrates the potential for early targeted interventions for MCI patients who are at increased risk of developing AD. The design of this model and the findings could also be used to guide further research evaluating the cost-effectiveness of other biomarkers used to identify MCI patients at increased risk of progression to AD.