The effects of an experimental geometry course on gender differences in standardized test performance were studied. The experimental approach involved: (1) reading new concepts and materials followed by class discussion; (2) relating applications to real-life situations; and (3) conducting a spiraling review of concepts throughout the year. Subjects were 810 students (49% were males, and 51% were females) in grades 9 through 12. Matched pairs, which each included one class using the University of Chicago Math Project textbook, "Geometry," and one class using a traditional textbook, were compared in terms of students' performance on a standardized test. The experimental classes reduced the gender gap by improving girls' test scores. On a content-specific test, girls in the experimental classes significantly reversed the trend of boys' dominance on applications and coordinates. Girls remained on a par with boys on items of visualization in three dimensions and on transformations. On the open-ended proof writing test, girls in the experimental classes performed as well as did boys. In traditional classes, a negligible gap remained, favoring boys. Four tables provide study data, and 10 graphs illustrate the trends. Sample test items are included. (SLD)