Abstract□□Objective: Assimilation model is a theoretical model concerning the change process of psychotherapy, which put forward by William B. Stiles and colleagues. The model suggested that, in successful psychotherapy, clients followed a regular developmental sequence of resolving the problematic experiences that they brought into treatment. The aim of this study was to verify the theoretical construct of Assimilation Model on the change process of successful psychotherapy, and to find the key point which might distinguish the cases of good and poor outcomes, by increasing the sample size to four cases with good outcome and four cases with poor outcome, and comparing the assimilation levels of these cases. Methods: Three research teams consisted of seven trained raters were engaged in the rating work, they used a qualitative research method of Assimilation Analysis to analysis the clients’ transcripts. Assimilation Analysis has been used to track the change process of clients’ problematic experiences in psychotherapy. Every investigator of the three research teams read and reread the transcripts of eight clients, tracked the major themes and described the voices, excerpted the relevant passages related the voices as the rating materials, then rated each passage according to the Assimilation of Problematic Experiences Scale (APES) and the Markers of Assimilation Manual. At last, every team reached consensus ratings of the excerpted passages. Results: The essential problems on those cases were discussed and reached a consensus. The internal coefficient consistency ICC (1,1) of the three teams ranged from 0.64 to 0.86; the ICC (1,K) ranged from 0.84 to 0.96. After the termination sessions, the mean score of the final assimilation (M = 5.50) and the mean score of the progress assimilation level (M = 4.25) for good outcome cases were higher than the poor outcome cases (with mean score of 2.13 and -0.13). The good outcome cases, along with the poor outcome cases, presented the process that therapeutic advances alternated with setbacks. However, the good outcome cases were able to maintain progress accompanying with setbacks, while the poor outcome cases failed to progress. All of the good outcome cases reached the Assimilation level 4 (understanding/insight), whereas the poor outcome cases failed to reach the level 4. Conclusions: Consistent with the former researches’ results, the findings supported the theoretical expectation that the reduction of symptom intensity were associated with higher levels of assimilation. All of the good outcome cases reached Assimilation Level 4, whereas none of the poor outcome cases did so. So the Assimilation Level 4 was the key point that distinguished the good and the poor outcome cases. However, in the view of all cases’ assimilation process, the therapeutic advances alternated with some setbacks，from these setbacks we could be made a conclusion that maybe some improvements on the model’s developmental sequence were needed. For example, the former linear developmental sequence could be revised as cycling up sequence.