本研究旨在發展一套可以測量人際特質與互動關係的人際行為量表，研究者參考相關理論自編題目，以中山大學及高雄師範大學學生209名，為預試樣本進行預試。項目分析選題後，並以MDS與三角函數方法，初步檢驗確認量表在圓形複合結構相符合。其次，以上述自編之人際行為量表為主要研究工具，並配合吳麗雲（民87）的人際互動評量表與王慶福（民84）的人際依附風格量表，編輯成大學生人際關係問卷，進行效度驗證的實徵研究。資料的取得主要透過立意取樣的方式，以人際行為量表與大學生人際關係研究問卷兩種工具，針對南區大學院校的學生，進行同性朋友間的調查，研究結果如下： 一、本量表具有可接受之內部一致性信度與重測信度。經多向度量尺分析與幾何學方法的驗證，證實三個人際層面的叢集順序皆符合理論構念。 二、人際行為量表人際間主動對人的行為與被動的人際反應之間，有顯著的關係存在；且此兩個人際層面間，都會受到過去內射自我概念的影響。 三、在配對者間，被動的人際反應，會受對方人際間主動對人的行為影響。且人際間兩個層面相關之組型成份，亦大致與互補原則及圓形複合結構相似的向度或象限位置相符合。 四、人際行為與人際困擾間，有顯著的效標關聯存在。 五、不同性別的大學生，其人際行為確有差異存在。 六、人際行為量表在人際間主動對人與內射自我概念兩個層面，可以區分部份人際依附類型間的差異。 基於以上之研究結果，研究者針對人際行為量表的使用與後續研究提出具體建議，供讀者參考。

　　　　　This thesis attempts to develop an Interpersonal Behavior Scale for the measurement of interpersonal features and interactions. Two hundred and nine college students of National Sun Yat-son University and National Kaohsiung Normal University serve as the pre-test samples. After item analysis, multidimensional scaling (MDS) and trigonometric method are used to examine how the scale matches in the circumplex. Using my own Interpersonal Behavior Scale as the main research tool, together with Wu Li-yun's Interpersonal Interaction Scale and Wang Ch'ing-fu's Interpersonal Attachment Scale,I edit a College Student Interpersonal Relationship Inventory for an empirical study of validity research. The material comes mainly through purpose sampling, using the Interpersonal Behavior Scale and College Student Interpersonal Relationship Inventory as the tools to examine the relationship between friends of the same sex in the southern Taiwan colleges. The valid questionnaires are respectively 125 for interpersonal behavior scale and 138 for college student interpersonal relationship inventory. The material is put into the computer and analyzed through SPSS according to the research hypotheses. The main statistic approaches are Canonical Correlation and two way MANOVA. What follows is the result: 1.In the scale, the coefficient of internal consistency and test-retest reliability coefficient can be tolerated. Multidimensional scaling and geometric approach prove that the three interpersonal relationship cluster orders correspond to theoretical construct. 2.In the scale, “action toward other” and “reaction to other” are obviously correlated, and both are influenced by “self-concepts”. Besides, in the composite of these two interactive dimensions, there is interpersonal flexibility, which shows that in the affiliation dimension an individual can be both friendly and hostile and in the autonomy dimension an individual can be both dominant and submit. 3.Between the matches, “reaction to other” is influenced by the other's “action toward other”. The composite of the two interactive interpersonal dimensions also approximately corresponds to the dimension or quadrant in the complementary principle and the circumplex. 4.Interpersonal behavior and interpersonal problems are obviously criterion related. 5.There are indeed differences in interpersonal behavior between two sexes of the college students. 6.As for “action toward other” and “self-concepts”, the Interpersonal Behavior Scale can distinguish between differences of interpersonal attachment types. Based on the above research result, the researcher purposes to the reader substantial suggestions as to the usage and development of the Interpersonal Behavior Scale.