Objective: To examine the diagnostic stability and the functional outcome of patients with early-onset psychosis (EOP) over a 2-year follow-up period. Methods: A total of 24 patients (18 males (75%) and 6 females (25%), mean age [plus or minus] SD: 15.7 [plus or minus] 1.6 years) with a first episode of EOP formed the sample. Psychotic symptoms were assessed using the Positive and Negative Syndrome Scale (PANSS). Social disability was measured with the Global Assessment of Functioning (GAF) disability scale. Diagnosis was assessed using the Kiddie-Sads-Present and Lifetime Version (K-SADS-PL), according to DSM-IV criteria. All diagnoses were re-assessed after 1 year and 2 years. Results: Schizophrenia had the highest prospective consistency (100% predictive value), while diagnostic stability was moderate for bipolar disorder (71.4%), and low for schizoaffective disorder (50%), schizophreniform and brief psychosis (50%), and psychosis not otherwise specified (NOS) (16.7%). The agreement between the baseline diagnoses and the 1-year follow-up diagnoses was 54.2%, whereas between the 1-year follow-up and the 2-year follow-up, it rose to 95.7%. Regardless of diagnosis, baseline negative symptoms were the only significant predictor of level of functioning at the 2-year follow-up (p = 0.010). Conclusion: Our results highlight the primacy of the first-year follow-up diagnosis in predicting diagnostic stability of EOP.