The content of selected major achievement tests is examined for bias in the frequency of references to the male and the female sex. The performance of males and females is contrasted on each item considered, and the relationships between item content and the performance differences is surveyed. The tests used in this study include the Metropolitan Achievement Test, the Iowa Tests of Basic Skills, the California Achievement Tests, and the Sequential Test of Educational Progress. Grades 2, 5, 8, and 10 are analyzed. Major findings include: (1) females perform better on items containing female references; (2) both males and females perform less well on items with a greater number of references; (3) consistent patterns occur regardless of whether the content variable studies is the total number of references or the total number of factors; (4) items with more than the average number of references do not differentiate ability levels within sex groups as sharply as do other items; and (5) both sexes are similar on such noncontent factors as omitting or rate-of-work. A Manual for Identifying Sex Bias Explicit in the Content of Test Items is appended. (Author/MV)