Only a small proportion of individuals with Mild Cognitive Impairment (MCI) will convert to dementia. Methods currently available to identify risk for conversion do not combine enough sensitivity and specificity, which is even more problematic in low-educated populations. Current guidelines suggest the use of combined markers for dementia to enhance the prediction accuracy of assessment methods. The present study adhered to this proposal and investigated the sensitivity and specificity of the electrophysiological component P300 and standard neuropsychological tests to assess patients with Alzheimer's disease (AD) and MCI recruited from a low-income country. The neuropsychological battery comprised tests of memory, attention, language, praxis and executive functions. The P300 was recorded using a classical visual odd-ball paradigm. Three variables were found to achieve sensitivity and specificity values above 80% (Immediate and Delayed recall of word list – CERAD – and the latency of P300) for both MCI and AD. When they entered the model together (i.e., combined approach) the sensitivity for MCI increased to 96% and the specificity remained high (80%). Our preliminary findings suggest that the combined use of sensitive neuropsychological tasks and the analysis of the P300 may offer a very useful method for the preclinical assessment of AD, particularly in populations with low socioeconomic and educational levels. Our results provide a platform and justification to employ more resources to convert P300 and related parameters into a biological marker for AD.