Examined the effect of sex and temperament differences on language-related differences in examination performance on written-answer and multiple-choice tests. 48 female and 25 male 3rd-yr medical students completed the Eysneck Personality Questionnaire and questions that required written or multiple-choice answers in an examination in psychology. There was a large sex difference in word production; variance of written answers loaded on 2 uncorrelated "linguistic" and "knowledge/cognition" components. Sex differences in temperament were also manifest but these contributed little to the observed differences in exam performance. There was no sex difference in multiple-choice performance, but women produced more words than did men and did better on written-answer questions. There were substantial and statistically significant correlations between word production and performance on both types of test.