Objectives: To evaluate the performance of the Peruvian version of the Rowland Universal Dementia Assessment Scale (RUDAS-PE) in discriminating between controls and patients with mild cognitive impairment (MCI) and dementia in an illiterate population with low-levels of education. Methods: We compared the cognitive performance of 187 elderly subjects who were illiterate (controls n = 60; MCI n = 64; dementia n = 63). Neuropsychological measures included the RUDAS-PE, Mini-Mental State Examination (MMSE), INECO Frontal Screening (IFS), and Pfeffer Functional Activities Questionnaire (PFAQ). The results were compared to a neuropsychological evaluation (gold standard), including use of Clinical Dementia Rating (CDR) scores. Results: We found a Cronbach's alpha was 0.65; Spearman's correlation coefficient was 0.79 (p < 0.01). The area under the receiver operating characteristics curve for the RUDAS to discriminate dementia from MCI was 98.0% with an optimal cut-off <19 (sensitivity 95%, specificity 97%); whereas, to differentiate MCI and controls was 98.0% with an optimal cut-off <23 (sensitivity 89%, specificity 93%). Conclusions: Based on its excellent psychometric properties, we find the RUDAS-PE suitable to aid in the opportune detection of dementia in a geriatric illiterate population with low-levels of education.