The success rate in predicting symptom change 2 years after brief dynamic psychotherapy increased 20% by adding quality of interpersonal relations from a psychodynamic evaluation interview to DSM-III diagnoses, Axis V and Axis II. The success rate in predicting dynamic change 4 years after therapy increased nearly 40% by adding the same variable. Selection criteria such as circumscribed focus, motivation and involvement in the evaluation interview were shown to be redundant as predictors.