大學入學考試中心2004年開始建置題庫，2006年開始從題庫抽取部份試題用於正式考試。本文以數學科為例，說明一個學科如何將多年的試題研究與題庫系統結合，從起始命題到最後考題評鑑的過程，來介紹大考中心題庫的發展。 數學科統整歷年研究，將「問題有意義、變化不求多、目標忌繁複、對象要區分、文字求通順、層次應分明、求證講清楚、難易需考量」等八大原則作為題庫計畫的準則。題庫計畫採個別命題、小組修題、專人審題方式進行，以達到落實研究成果、擴大命題人才、提昇與穩定試題品質、有效管理與保全試題的目的，並採用「因素指標模式」取代傳統預試，以預估難易度。考後結果呈現此難易度預估方法確實可行，且多數題庫題對高低分組具鑑別力，亦讓整份試卷結構更完備。但未來需持續進行掌控難易度、測驗目標等質性指標的研究，並輔以IRT（Item Response Theory）進行試題的檢測。有些未被採用的題庫題已發展成學科知能量表，且施測兩年，達到題庫多功能使用的目的。建議未來仍須強化闈外組卷的功能，減少題庫題的耗損，並發展多種測驗。

The item bank of CEEC has been established since 2004, and some items have been used in the Department Required Test (DRT) since 2006. This paper uses the Mathematics test as an example to explain how test writers of a scholastic subject can make use of research on test items and item bank system in the process from test construction to test evaluation. By doing so, the development of the CEEC Item Bank will be introduced. Based upon previous research, the Math test item bank has been planned on the basis of eight principles: meaningful items, fewer variations, simple objectives, differentiated test candidates, readable language, distinct levels, clear solution, and taking difficulty level into consideration. The purpose of the plan was to promote item quality, stabilize item difficulty indexes, increase the number of item writers, and ensure good item management and item security. A Factor-Indexed Model was used to replace the traditional pre-testing approach to predict test difficulty. The DRT test result analysis showed that this new index was practical and reliable, being able to discriminate high-achievement and low-achievement students. However, more qualitative research is needed for controlling difficulty levels and clarifying test purposes, with the assistance of IRT for item evaluation. As for items not used in the DRT, they have been used in other kinds of tests so as to multiply the uses of the item bank. It was proposed that the functions of the item bank be further strengthened and variety of tests be developed.