Objectives: Although people with amnestic mild cognitive impairment (aMCI) benefit from cerebrolysin treatment, some still develop dementia. The aim of the current study was to identify most informative clinical assessment tests to predict the therapy efficacy in aMCI subjects treated with cerebrolysin. Methods: We studied patients with amnestic mild cognitive impairment (aMCI; n = 53) who had regular neurocognitive and clinical psychiatric assessments and were treated with cerebrolysin i.v. infusions 20 × 30 ml twice a year over three years period. Data were analyzed using non-parametric statistics, cluster and linear discriminant analyses. Results: Combined mathematical modeling enabled to predict cognitive decline from aMCI to dementia in the cerebrolysin-treated patients based on their initial neurocognitive assessment scores. We identified a "dementia risk group" with fast cognitive decline (i.e. low efficacy of the treatment). Lower baseline scores in the Mattis Dementia Rating Scale Memory subtest, Mini-Mental State Examination (MMSE), 10 word list immediate recall, and Frontal Assessment Battery tests when accompanied by higher depression score (Hamilton Depression Rating Scale) suggest poor prognosis for aMCI patients treated with cerebrolysin. Changes in scores on the MMSE, Boston naming test, Digit span forward, and Wechsler scale subtest "Categorical associations" during the treatment course are more characteristic for patients who convert to dementia than their initial scores. Conclusions: aMCI subjects treated with cerebrolysin with lower baseline cognitive functioning and subclinical depres-I. S. Boksha et al. 2582 sion have poor prognosis in terms of converting to dementia. Changes in the MMSE, Boston naming test, Digit span forward, and Wechsler scale subtest "Categorical associations" scores during the treatment course are more informative to identify patients who will develop dementia than their initial scores.