The purpose of this study was to validate a reduced version (15 items) of the Boston Naming Test (BNT) in a sample of 78 low-educational elderly persons with or without dementia, as determined by independent assessment with a battery of cognitive tests. The reduced version was found to be equivalent to the complete BNT, and to have criterion validity with respect to other measures of dementia. We conclude that the reduced version is a useful instrument for assessing patients who require shorter testing methods because of severe cognitive deterioration or their low level of education.