In the present study of 203 patients in psychodynamic group psychotherapy, we explore associations between patient and therapist global retrospective outcome evaluations (ROE), and pre-post-treatment changes on the Symptom Check List 90 Revised (SCL-90-R) and non-symptomatic focus of therapy. There were no significant associations between ROE, diagnoses and demographic variables, and pre-treatment SCL-90-R associations were negligible (less than 4% of overlapping variance). SCL-90-R subscale improvement expressed as residual gain score explained the overall largest percent of variance in both patients and therapists (44% and 25%, respectively) when compared with raw difference scores (32% and 18%), and percent reduction from baseline (34% and 23%). Moreover, ROE/end-state adjustment associations were substantial (42% and 24%). Therapists' evaluation of change showed the strongest association with improvement in non-symptomatic focus of therapy, while patients' evaluation had the strongest association with improvement in SCL-Depression. It is concluded that retrospective evaluations reflect changes related to treatment. However, unexplained variance may be independent of symptomatic state, and associated with personality factors or domains not captured by standard questionnaires.