Naturalistic psychotherapy effect studies commonly report effect sizes for the total sample. However, a previous study of SCL-90 Global Severity Index (GSI) improvement in a large outpatient sample used a cluster analytic strategy and reported clinical relevant outcome trajectories that could be grouped into early within-treatment improvement, late improvement in the follow-up period, and deteriorating patients with slight improvement that was lost at follow-up. We explore GSI outcome trajectories and clinical significant change in a sample of 320 patients at a public psychiatric outpatient psychodynamic group therapy unit, the majority with anxiety, personality, and mood disorders. The study revealed large discharge and follow-up effect sizes but more than one third of the patients were without measurable improvement. The major clusters described above were confirmed, and revealed unique clinical and socio-demographic characteristics. Late improvers, as compared with early improvers, were characterized by anxiety symptoms and lack of network support after controlling for GSI at admission. Similarly, deteriorating patients had longer duration of illness and less favourable social characteristics compared with the other two groups. Early improving patients were less likely to have participated in short-term groups, and only one third participated in additional treatment compared with more than 69% of the other patients. Severe and socially affected psychiatric patients, and patients with anxiety and agoraphobic symptoms may be less optimally treated in short-term time limited psychodynamic groups. There is an obvious need for diversity of treatment offers, better integration of psycho-social treatment components, and long-term open ended treatment.