Objectives To describe, in a naturalistic setting, the impact of the early use of LAI-AP on functional outcomes of early psychosis patients as compared to oral antipsychotics (OAP). Methods Longitudinal prospective 3-year naturalistic study of all consecutive admissions (n = 416) to two Early intervention services (EIS) for psychosis comparing baseline characteristics and the evolution of global functioning, occupation (work and studies), and living arrangements autonomy according to the route of administration of the antipsychotic medication. The cohort was divided into four groups: LAI-AP first (started on LAI-AP and later received OAP), OAP first, LAI-AP only, and OAP only. Results Global assessment of functioning (GAF) improved in all groups, but our mixed-effect model did not show any significant association between the route of administration and the GAF outcome. The LAI-AP only group was significantly less likely to have extreme residential instability at 3 years than the other groups despite its highest proportion of homeless youth and their poor prognostic factors at baseline. Conclusions Our naturalistic study suggests a significant protective effect of LAI-AP on extreme residential instability for the most vulnerable patients, but no impact of the first AP administration route on other functional outcomes was observed at 3 years of follow-up. Key points Long-acting injectable antipsychotics seem promising to avoid extreme residential instability in early psychosis. Global assessment of functioning (GAF) improved in all groups. There was no significant association between the first route of administration and global functionning.