Background <p>Soteria houses are small-scaled residential settings for the inpatient treatment of acute, early episode psychosis, in which recovery is pursued by offering a calming, normalizing environment and being present. The current study aims to compare the longitudinal effect of Soteria on Personal Recovery (PR) with community-based care as usual (CAU) in the Netherlands.</p> Methods <p>The two year course of PR both on individual level and mean group level was followed in patients receiving Soteria (N = 28) or CAU (N = 94). PR was assessed with the I.ROC. As potential confounders baseline scores of psychosis symptom severity (PANSS-R), impairment in functioning (WHODAS), internalized stigma (ISMI), and hospital admissions were assessed. Individual change scores and multilevel analyses were used for comparing the course of I.ROC scores between Soteria and CAU.</p> Results <p>I.ROC scores in the Soteria condition improved more than in CAU, both on individual and mean group level. Multilevel analysis revealed 3 points higher I.ROC scores for the Soteria group after two years, however the effect became non-significant after correcting for baseline measures of symptom severity and impairment in functioning.</p> Conclusions <p>Although concepts like Soteria show promise to transform inpatient services for acute early psychosis in a recovery-oriented manner, we were unable to demonstrate robust effects of Soteria compared to CAU after 2 years, when accounting for between‐group differences. Potential explanations and recommendations for future research are discussed, illustrating the need for more research into ways to promote PR from early episode psychosis in inpatient care.</p>