Computer adaptive testing is the study of scoring tests and questions based on assumptions concerning the mathematical relationship between examinees' ability and the examinees' responses. Adaptive student tests, which are based on item response theory (IRT), have many advantages over conventional tests. We use the least square method, a well-known statistical method, to reach an estimation of the IRT questions' parameters. Our major goal is to minimize the number of questions in the adaptive test in order to reach the final level of the students' ability by modifying the equation of estimation of the student ability level. This work is a follow-up on Al-A'ali (2007). We consider new factors, namely, initial student ability, subject difficulty, number of exercises covered by the teacher, and number of lessons covered by the teacher. We compared our conventional exam results with the calculated adaptive results and used them to determine IRT parameters. We developed the IRT formula of estimating student ability level and had positive results in minimizing the number of questions in the adaptive tests. Our method can be applied to any subject and to school and college levels alike. (Contains 14 figures and 3 tables.)