This study is an attempt to test and examine hypothesis 1 (a),(b) and 3 in the experimental hypotheses by G. T. Barrett-Lennard. By their hypothesis 1 (a), the relationship variables as measured after interviews at initiation of therapy, were assumed to be positively correlated with the therapeutic change during the period of therapy. Moreover, these associations were predicted to be stronger when the relationship qualities were measured from client perceptions than when they were measured from therapist perceptions. However converse results have come out in this research. The hypothesis 1 (b) predicted that the relationship as measured from client or therapist perceptions at termination of therapy would also be positively associated with change. For this hypothesis this researcher has found that correlation at termination of therapy was stronger in the case of therapist perceptions than in client perceptions. So far as their hypothesis 3 is concerned, the same outcome has been produced: that is, combined relationship perception total scores of the two participants (clients and therapists) were more associated with therapeutic change than scores of either one of them (the clients or therapists). The fact that could be stated in common after examinations of these three hypotheses in this correlation between the variable of empathic understanding and the therapeutic improvement were greatest. From these investigations, the next questions will be presented: 1) The generalization of client perceptions 2) The limitation of perceived relationship inventory (1) In the procedure of the generalization of client perceptions by Barrett-Lennard, the response of the differential therapists related to the identical clients would be responsible for the different perceptions of the clients and, hence, for differences in therapy outcome. In this paper, for identical therapist attitude related to different clients, they express different responses. This difference would come out as the variable of empathic understanding in therapist perceptions. (2) Variables of “self-congruence” and “unconditionality of regard” in therapist perceptions secure the stability of attitude necessary for therapists in spite of therapeutic change. On the other hand, client perceptions are not responses for the three variables (congruence, empathy, and unconditionality) but these are understandable as reflection upon the inner state with dependency, hostility, disturbancy.