Cognitive function was assessed in 930 elderly Nigerians living in an urban community by trained medical students who were blinded to the educational status of the subjects. They used a pretested and validated questionnaire. The subjects were later evaluated for the presence of dementia according to DSM-IIIR criteria by neurologists who then determined the educational status of the subjects. There were 361 males and 569 females with a mean age of 58.5 years (SD = 14.8). One hundred and sixty-one subjects (17%) with at least 6 years of education performed better on all aspects of cognition assessed (p < 0.001). The differences in performance were statistically significant on assessment of general knowledge, intelligence, abstract thinking, depth of information and immediate recall (p < 0.001) but not for calculation, orientation, attention and language comprehension. Our findings confirm the confounding effect of education on cognitive assessment and suggest the possibility of developing a short culture-free and educationally unbiased instrument for assessing cognitive functions in Nigerians.