Treatment effects in trials with trauma-affected refugees vary considerably between studies, but the variability in outcome between individual patients is often overlooked. Consequently, we know little about why some patients benefit more from treatment than others. The aim of the study was therefore to identify predictors of treatment outcome for refugees with Posttraumatic Stress Disorder (PTSD). Data was derived from two randomised trials including 321 refugees, who had all participated in a 6-7 months bio-psycho-social treatment programme. Outcome measures were the Harvard Trauma Questionnaire (PTSD, self-rating), Hopkins Symptom Checklist-25 (depression and anxiety, self-rating) and Hamilton Depression and Anxiety rating scales (observer-ratings). Using hierarchical regressions models, associations were analysed between pre- to post treatment score changes (dependent variable) and a range of variables including sociodemographics, pre-migration trauma, post-migratory stressors, baseline symptom scores and level of functioning. A high baseline score (=more symptoms) and a high level of functioning were found to be associated with improvement on all ratings. Additionally, the following variables were associated with symptom improvement on at least one outcome measure: short time in host country, full time occupation, young age and status as family reunified (in contrast to refugee status). Being Muslim was inversely correlated with improvement. Translated self-ratings were used, which could impact reliability. These results call for screening and early interventions for arriving refugees. For clinical populations, level of functioning should be included in assessments of refugees, to possibly begin stratifying samples to different interventions based on their likelihood of responding. Copyright © 2020 Elsevier B.V. All rights reserved.