本研究目的在以試題反應理論為基礎，建立一量尺分數，在兼顧學理的嚴謹性和使用方便性上，以更為合理和公平方式估計受試者能力值。本研究以試題反應理論之單參數對數型模式，試題參數估計採邊際最大概似估計（MML），使用EM演算法，受試者能力值估計採加權概似估計法（WLE）。實證資料分析結果顯示，99～103年歷屆各學院間新生的英文能力呈顯著性差異，100年入學之新生英文能力最佳，之後逐年呈現顯著性下滑趨勢；管理學院新生英文能力最佳、航輪學院次之、海工學院再次之、水圈學院的新生英文能力最差；歷屆測驗試題參數內部一致性檢定介於0.793～0.869之間，顯示由外語教育中心每年所舉辦的「大一英文能力分級會考」歷屆測驗試題，具有穩健的試題參數指標，可藉由IRT單參數對數型模式精確、有效的估計受試者能力值，經由量尺化過程，明確描述受試者能力值之變化，提供給授課教師在教學教材選用與教學上的參考，以供實務應用。

The purpose of this paper is to use Item Response Theory (IRT) to develop a scaled score gained from years of English Placement Test (EPT) of National Kaoshiung Marine University (NKMU) students. The proposed IRT considers both strictness and convenience of conducting a fair and more reasonable method to calculate the performance of EPT for NKMU students. In this research, data are built up based on IRT single-parameter logarithm model and item parameter estimate is acquired by Marginal Maximum Likelihood (MML) method. In consequence, by mean of EM algorithm we achieve an outcome which demonstrates the performance of EPT test takers calculated by Weighted-Likelihood-Estimation (WLE). In a result, 2011 freshmen have the best performance in EPT while there are notable decline trend afterward. Besides, among colleges Management College comes in the first place, Marine Engineering College in the second, Ocean Engineering College in the third and Hydrosphere Science College in the last place. Furthermore, we obtain consistent output values (0.793 ~ 0.869) of parameters from each EPT test, which shows EPT held by the Foreign Language Center in NKMU each year can affirmatively provide stable and reliable function. Moreover, with IRT mode, the variation of testers’ English capability can be clearly described and tracked through its accurate, effective assessment and scaling process. Consequently, teaching materials and skills can be precisely chosen for improving their English capability.