OBJECTIVE The objective of this study was to study baseline clinical and biological predictors of 2-year outcome in a cohort of children and adolescents with a first episode of psychosis. METHODS Standard instruments were used to evaluate symptoms and functioning in 110 children and adolescents (mean age = 15.47 years) with first episode of psychosis at admission (between 2003 and 2005) and after 2-year follow-up. Clinical assessments included diagnostic assessment to yield DSM-IV diagnosis, developmental, premorbid, and past-year data, together with structural neuroimaging and other biological parameters (genetics and oxidative stress). Eighty-three subjects had assessments at baseline (including the Strauss-Carpenter Outcome Scale [SCOS]) and at 2-year follow-up. Association and multistep regression analyses were conducted to show correlates and predictors of primary outcome measures: functional outcome (Children's Global Assessment Scale [CGAS]), improvement (CGAS change), and primary negative symptoms (Proxy for the Deficit Syndrome Scale). RESULTS The SCOS predicted 27.46% (P < .001) of the variance in CGAS score at 2 years. Baseline severity (measured by CGAS) predicted 30.9% (P < .001) of CGAS improvement after 2 years, and SCOS total score predicted an added 24.1% (P < .001). A diagnosis of nonaffective psychosis, primary negative symptoms, and less white matter at baseline predicted more primary negative symptoms at follow-up. The prediction of functional outcome was not increased by genetic, oxidative stress, or neurostructural markers. CONCLUSIONS Baseline clinical assessments have a better predictive value than biological assessments for 2-year follow-up functioning of children and adolescents with a first episode of psychosis. Patients with primary negative symptoms at baseline continue to have negative symptoms 2 years later, and neurostructural markers predict these. Clinicians must still rely on clinical variables to judge the functional prognosis of early-onset first psychotic episodes.