研究目的：本研究釐清不同成人依戀風格在羅夏克墨漬測驗上的表現，並檢證Fonagy的精神分析理論，進而反思臨床心理學上的應用。 研究程序：研究期間，共招募到302位大學部學生填寫《親密關係經驗量表—關係結構版》。而後，根據量表信、效度分析後，篩選出高焦慮依戀分數、高逃避依戀分數及兩分數均低之控制組各25人，邀請其參與羅夏克墨漬測驗。最終收取焦慮依戀組12人、逃避依戀組12人、安全依戀控制組15人進行資料分析。 研究結果：認知及投入向度方面，焦慮依戀者與逃避依戀者在F%顯著高於安全依戀者；同時，焦慮依戀者在Complexity、Sy、Blend、M及MC 表現亦差於安全依戀者，答題使用範圍則顯示Dd%較高、W%較低。情緒指標方面，焦慮依戀者在V變數上顯著高於其他兩組。人際互動方面，焦慮依戀者的SumH及H顯著低於安全依戀者。 研究貢獻：本研究初步支持Fonagy精神分析論於不安全依戀者的認知與心智化、情緒控制與人際表徵分化等證據。此一結果亦可增加臨床實務中對不安全依戀者的衡鑑遞增效度，並探究其背後的心理病理機制。 / Objective: When attachment comes to assessment, many clinicians would utilize self-report inventories to gather information. However, different attachment styles, in fact, represent diverse behavior patterns; thus, using instruments adopting multi-rating techniques such as the Rorschach would be ideal. Based on Fonagy’s psychodynamic theory and past Rorschach results, we predict participants with different attachment styles will have different pattern in Rorschach task. Methods: We conducted a double-blind research design and recruited 302 volunteers. All participants were required to fill out the Experiences in Close Relationship - Relationship Structure inventory, a measurement used to classify the adult attachment styles. The top 7% of highest scoring participants on the Anxiety Dimension were assigned to the Anxiety Group(ANG), and the same strategy was applied to those in the Avoidance Group(AVG). The Control Group(CG) consisted of 7% of the randomly-selected individuals who scored below the means of both Anxiety and Avoidance Dimensions. Participants in these three groups were invited and encouraged to take the Rorschach test according to the R-PAS. Results: Current research collected 39 participants Rorschach protocols (n = 12 for both the ANG and ACG; n = 15 for the CG). Results shows that both ANG and AVG have higher F% then CG, and ANG have lower Complexity, Sy, Blend, M, W%, MC, SumH and H then CG. ANG have higher V then AVG and CG. Conclusion: Current results partially supported Fonagy’s psychodynamic theory. Results also support using multi-rating tasks to increase assessment validity when evaluating different attachment styles in clinical settings.