It is known that speed and accuracy in recognizing words are constrained by the frequency of occurrence of these words ("frequency effect"). This study examines the relationship between educational level and the word frequency effect. We postulated that individual exposure to words that are rated lower in frequency tables should be greater among subjects with higher education and therefore hypothesized that the magnitude of the frequency effect should not be as marked within such a population as among subjects with a lesser educational level. A total of 40 neurologically healthy adults, half with an average of 18 years of formal education and the other half with an average of 11 years, participated in a lexical decision experiment. Results confirmed our hypothesis; that is, significant frequency effects on reaction times were obtained in both groups but this effect was of greater magnitude for the less educated as opposed to the more educated subgroup. The significance of this finding is discussed by reference to current models of word recognition.