Objective: The current study investigates the use the Mini-Cog in the detection of possible dementia. Also, explore the relationship among the independent variables MMSE and the CERAD ten wordlist delayed recall with the scores of the Mini-Cog as the dependent variable. Method: The current study is a secondary analysis with an exploratory-descriptive design. The total sample consists of 114 participants (n = 83 females and n = 31 males) recruited from residential federal funded project homes for the elderly known as “Égidas” and Elderly Care Centers municipal government funded projects. The age range was 55–88 years (M = 71.81, SD = 7.81). The administrated test, were: Mini-Cog, MMSE, and the CERAD ten wordlist delayed recall. The Mini-Cog has a sensitivity ranging from 76–99% and specificity ranging from 89–93% with 95% confidence interval for the Puerto Rican population from 60 years and over (González-Viruet, Pérez-Mojica y Rodríguez-Gómez, 2014). Results: The Mini-Cog identified 46.8% of the sample with possible dementia, while the MMSE identified 49.1% (score < 26) and the CERAD ten wordlist delayed recall a 61.9% (score < 5). All screening instruments result in a significant direct and low moderate correlation (p < 0.05) for academic level except the wordlist delayed recall. The multiple regression equation explain 60% of the Mini-Cog scores variance, identifying the MMSE score and the CERAD ten wordlist delayed recall as the best predictors. Conclusion: This results support the use of the Mini-Cog as a screening tool for the possible detection of dementia. The use this assessment tool can help the health system to make a quick, accurate and reliable screening for the elderly population.