Differential functioning of males and females on achievement test items was studied in a sample of 110,000 students in kindergarten through grade 12. First, item parameters and ability estimates for these parameters were obtained from LOGIST computer program runs for the entire group. Predicted performance for each item for each group was separately determined, and expected performance was then subtracted from observed performance for the item. Items with extreme values were examined for content bias. Females had higher mean scale scores on all but three of 72 tests in the basic battery. Differences favoring females increased up to grade 8 and then diminished in high school. Variations of scale scores for females were smaller than those for males. The number of items biased in favor of females was less than the number biased in favor of males. Until such time as these results are better explained, test constructors are advised to continue carefully balancing the number of male and female names and actors in passages and items. Test data are summarized in attached tables. (SLD)