Background: The increase in life expectancy and proportion of elderly in the population is causing an increase in dementia prevalence rates. The correct, early diagnosis of dementia is very important to clinical treatment and to improved prognosis. Therefore, it is necessary to adapt and develop assessment tools for the differential diagnosis between pathological and normal aging processes. Objective: Assess the psychometric properties and the factorial structure of a neuropsychological protocol used in geriatric assessment. Method: Subjects (n = 69) with heterogeneous cognitive complaints were assessed at the Centro de Geriatria e Gerontologia at the Hospital das Clínicas da Universidade Federal de Minas Gerais using a protocol composed of the Mini-Mental State Examination, Clock Drawing, Corsi Blocks, Verbal Fluency, Digit Span and Token Test. Statistical analyses included factorial analyses of test results, Pearson’s correlation between obtained factor, age, years of formal education and clinical dementia rating and area under the ROC curve. Results: The factorial analyses of test scores showed a general representative factor that had moderate and significant association with CDR (r = -0.672; p < 0.001) and years of formal education (r = 0.455; p < 0.001), respectively. This factor had weaker and less significant correlation with age (r = -0.282; p < 0.05). Discussion: These results point to the protocol’s good construct and criteria validity in assessing cognitive decline in the elderly. Future works concerning applicability and populational norms are needed to improve the clinical use of this assessment protocol.