The objectives of this study were to develop, implement, and evaluate the year long project, Microcomputer Adaptive Testing High-Risk Urban Students (MATH-R-US). The project produced diagnostic software to meet the following criteria: (1) help students obtain high school mathematics credit needed for graduation; (2) motivate students to learn mathematics; (3) account for erratic student attendance; and (4) use computer adaptive testing as an integral part of the program. The project was used for an entire school year by a class in an urban high school with an at-risk predominantly black population and a high rate of absenteeism. The tests, which accept generative responses rather than multiple choice answers, were administered once a week in the school's computer lab. The results of each test were saved and practice sheets, with answer keys, were generated for the missed objectives. The program was evaluated to improve implementation and furnish descriptive data to the classroom teachers and school administrators. It was found that the program generated intense student competition to see who could get the most hamburger graphics--which appeared on the screen when students completed a test with 100% accuracy--in an hour. Both males and female students expressed positive attitudes about the course components, but female responses reflected more confidence in their own abilities. Computer math test scores indicated a consistent improvement on retesting of a topic, with 23 perfect scores on 43 retests. A discussion of the implications of this study concludes the paper. (2 tables, 22 references) (BBM)