This study aimed to determine whether performance on tests of crystallized intelligence is affected by gender and to ascertain whether differential item parameters could account for the gender disparities. The sample comprised 1.191 individuals (55% women) between the ages of 16 and 77 years old (M=22; SD=9.5). The participants were primarily college students (58.3%) living in four Brazilian states. Four verbal tests measuring crystallized intelligence (vocabulary, synonyms, antonyms and verbal analogies) were constructed and administered in a group setting. An analysis of variance revealed no significant differences in the overall performance between men and women. However, a differential item functioning analysis indicated significant differences on 8.7% of the items, which indicates the existence of gender bias. Because bias can limit women’s access to social opportunities, the results obtained indicate the importance of reducing item bias in cognitive measures to ensure the accuracy of test results.