The goal of this study is trying to build an item bank of computer adaptive diagnostic test. A paper-based pre-test was developed based on experts’ structure. The pre-test data was analyzed based on ordering theory and item relational structure analysis to get different item structure. Experts’ structure and item structure were implemented in computerized adaptive diagnostic test. Four adaptive testing algorithms based on structure theories are proposed in this paper. The results show that the adaptive testing algorithms based on structure theory and experts’ structure is robust and has the best performances in both testing speed and prediction accuracy.