Objective: To assess the role of brief neuropsychological assessments in prediction and identification of Alzheimer's disease (AD) pathology and progression to AD dementia.Method: 255 adults (40-81 years) with self-reported cognitive decline underwent baseline and two-year follow-up clinical assessment, including a brief neuropsychological screening and lumbar puncture. Five different mild cognitive impairment (MCI) algorithms were applied on baseline cognitive test results: one conventional, three amnestic (lenient, stringent, multi-domain), and one comprehensive criterion. We compared predictive and diagnostic accuracy of these MCI criteria by performing logistic regression analyses and calculating diagnostic accuracy measures for two-year outcomes of: 1) clinical diagnosis of AD dementia, and 2) cerebrospinal fluid biomarkers in the Alzheimer's continuum. Results: The lenient amnestic MCI criterion showed the largest effect size for predicting progression to AD dementia (OR=13.762, 99% CI=1.969-96.194, p=.001), and AD biomarkers OR=4.855, 99% CI=1.974-11.924, p<.001) after two years. This criterion was sensitive for progression to dementia (sensitivity=92.0%, specificity=54.8%, LR+=2.03, LR-=0.15) and showed the highest overall diagnostic accuracy for AD biomarkers (sensitivity =72.7%, specificity=59.1%, LR+=1.78, LR-=0.46). The multi-domain amnestic MCI criterion produced the highest specificity for dementia (sensitivity=76.0%, specificity=73.0%, LR+=2.82, LR-=0.33) and AD biomarkers (sensitivity=46.8% specificity=70.9% LR+=1.61, LR-=0.75). Conclusions: Defining MCI using a brief neuropsychological battery provided limited accuracy for progression to AD dementia and CSF Aβ. The lenient amnestic MCI criterion identified the highest number of individuals who progressed to clinical AD or showed biomarker pathology, but this approach included a substantial number of false positives.