In general, the prognostic relevance of a working alliance for both the course and the outcome of psychotherapy is no longer a matter of dispute. Since a working alliance is conceived as an interactional variable, the results must be interpreted on the basis of several important dimensions: the perspective of the investigator and the time structure during the course of therapy. Using the data available from the Berlin Psychotherapy Study (Rudolf 1991), we investigated for the 238 patients (inpatients and outpatients who received psychoanalytically oriented therapy in the study) the prognostic relevance of diagnostic and therapy-related working alliance variables for the various outcomes investigated. A complex correlation-statistical method (latent-trait-model) was used to consider the investigator perspective with regard to the working alliance and the outcome as well as the time structure and the diagnostic variables. Our results emphasize the relevance of the therapist's perspective beginning with the diagnostic indication, the working alliance and finally the outcome. The patient's perspective, however, seems to be less relevant. The results are presented and discussed in relation to the relevant literature. One result we have found as a further important dimension is that the kind of therapy carried out must be taken into consideration. This implies that as in our study for all phases of the investigation therapy specific instruments must be applied. In this regard our results also contribute to the validation of the instrument used to measure the working alliance TAB.