國防部為確保軍事院校能有效運用有限的國防預算，以達成辦學成效，需要一套合理客觀的績效評估方式。研究目的在說明如何能以資料包絡分析法( Data Envelopment Analysis， DEA )作為一個軍事院校辦學績效評估管理方法，使國防部能有效地對軍事院校作好績效管理。本研究以國防管理學院為例，使用85-90年度教育資料：4個投入變數(教師人數、人員維持費、作業維持費及軍事投資)及4個產出變數(畢業學生人數、期刊篇數、其他著作及研究收入)，以DEA與窗口分析 (window analysis) 來評估該院六個系所在不同時期的辦學 續效。實證分析發現：(1) 資源所有較佳平均純技術效率(99.07% )；(2 ) 會計系有較佳平均規模效率 (99.71 % )；(3) 會計系所與資訊系所最適生產規模之比例最高 (91.67%)；(4 ) 資訊系所有較佳平均成本效率 (99.90% )；( 5 ) 經辦學績效目標改善分析後，可發現減少人數、人員維持費作、作業維持費與增加學生畢業人數與教師其他著作篇數，是無效率系所共同努力的目標；(6 ) 由樣本系所辦學效率與成本效率之相關分析中，本研究無法獲得強力統計支持，推論出一個有高辦學效率系所同時亦會有高成本效率；(7) 當樣本數較少時，將DEA與FDH作一比較後，FDH發覺有效率的樣本系所增多，顯示FDH較不嚴謹，無法找出真正有效率單位。最後，本研究提出管理意涵與改善建議，以提供國防部及軍事院校作為績效管理之參考。

In order to ensure its military academies effectively use limited defense budgets in higher education, the Ministry of Defense needs a managerial tool for monitoring the performance of military academies. This purpose of this paper is to demonstrate the use of Data Envelopment Analysis (DEA) in evaluating the performance of academic departments at National Defense Management College over 1996-2001 in conjunction with window analysis. Four inputs (number of teaching staff, personnel salaries, operating expenditures, capital investments) and four outputs (number of undergraduates and postgraduates graduated, number of journal paper published, edited books, and research income) are used for this study. Results of this study show: (1) the Graduate School of Resources Management has the highest mean pure technical efficiency score of 99.07%; (2) the Department and Graduate School of Accounting has the highest mean scale efficiency score of 99.71 %; (3) the Department and Graduate School of Accounting and the Department and Graduate School of Information Management both have the same highest proportion (91.67%) of DMUs experienced constant returns to scale under four window analyses over six years; (4) the Department of and Graduate School of Information Management has the highest mean cost-efficiency score of 99.90%; (5) there are rooms for inefficient DMUs to make improvements in reduction of the number of teachers, personnel and operating expenditures, and in increase of the number of student graduated and publications by teachers; (6) a department that has high technical efficiency doest not mean that it also has cost-efficiency; and (7) FDH identified more efficient DMUs in comparison with DEA, this may imply that FDH has a weak discrimination power of identifying efficient DMUs under a small amount of samples. Finally, managerial implications for the Ministry and the College are also presented.