The primary objective of this study was to investigate overall patterns of gender differences and similarities of test performance in mathematics. To achieve that objective, observed test scores on the Delaware standards-based assessment were analyzed to examine: (1) gender differences and similarities across grades 3, 5, 8 and 10 over 2 years; (2) the existence and magnitude of the effects of item format on gender performance; and (3) the existence and magnitude of the effects of test construct, content standards, and cognitive processes on gender performance. The similarities of the overall test performance and distributions of test scores between females and males indicate no gender differences on the Delaware Student Testing Program (DSTP) mathematics assessment and the mathematics assessment of the Stanford Achievement Series, ninth edition (SAT9), also given to these students. Evidence from these analyses suggests that gender differences do exist when the significant differences found in variability are combined with mean differences among students scoring at the top and bottom 10% of the DSTP and the SAT9. The results of the data analysis indicate a slight, but consistent format effect across grades, with males performing better than females on multiple-choice items and females performing better on extended constructed response items. Unlike the results of early studies, no clear patterns of test performance between females and males have been found for the content and cognitive categories of mathematics. Five appendixes contain a discussion of the mathematics content standards, definitions of the cognitive processes of the DSTP mathematics assessment; a discussion of item format, sample items with scoring rubrics, and mathematical references sheets for the assessment. (Contains 7 tables and 39 references.) (SLD)