The multistage alpha-stratified computerized adaptive testing (CAT) design advocated a new philosophy of pool management and item selection using low discriminating items first. It has been demonstrated through simulation studies to be effective both in reducing item overlap rate and enhancing pool utilization with certain pool types. Based on this design, two extended stratification methods have been proposed to deal with practical issues of content constraints and correlation between difficulty and discrimination parameters respectively. These stratification designs on their own do not automatically meet content requirements. Instead, they need to be modified or have to work together with some content balancing methods in order to satisfy all content constraints. This study aimed to investigate through simulation whether there was any effect due to the factors of content balancing method and stratification design. The three stratification designs were examined together with three well-developed content balancing methods, under the practical constraint of strict content specifications. The performance of each of the nine combinations was evaluated in light of item security, measurement efficiency, and pool utilization. Results indicate that there is no interaction effect between the two factors, but main effect does exist in the content balancing method on item security and pool utilization, as well as for stratification design on item usage. (Contains 4 figures and 19 references.) (Author/SLD)