Objectives: (1) Assessment of diagnostic stability of psychotic disorders or psychotic mood disorders from 6 weeks to 18 months after initiation of treatment in a representative first-episode psychosis (FEP) sample. (2) Comparison between those patients who shifted from DSM-IV schizophreniform disorder to schizophrenia or schizoaffective disorder and those whose diagnosis of schizophreniform disorder remained stable. Method: The Early Psychosis Prevention and Intervention Centre (EPPIC) in Australia admitted 786 FEP patients from January 1998 to December 2000. Data were collected from patients' medical records (MRs) using a standardized questionnaire. Seven hundred four MRs were available, 36 of which were excluded owing to nonpsychotic diagnoses or a psychotic disorder due to a general medical condition. Of the remaining 668 patients, 176 (26.3%) were lost to follow-up. Four hundred ninety-two subjects were analyzed. Strategies to assure validity and reliability of diagnoses were applied. Results: The same diagnosis was made at baseline (<= 6 weeks after admission into EPPIC) and 18 months for 69.9% of the patients. Among the most consistent diagnoses were schizophrenia (97.3%), schizoaffective disorder (94.1%), and bipolar disorder (83.2%); the least stable, as expected, was schizophreniform disorder (40.0%). In subjects with schizophreniform disorder at baseline, the best predictors of a shift from schizophreniform disorder to schizophrenia or schizoaffective disorder were a higher baseline Clinical Global Impressions-Severity of Illness scale score and lower premorbid Global Assessment of Functioning score, although the variance accounted for was small (R-2 = .07). Conclusions: A longitudinally based diagnostic process in FEP samples is needed, especially in schizophreniform disorder and bipolar disorder. However, a thorough initial assessment of patient and family by a specialized team of investigators regarding the kind and duration of patient symptoms may lead to high diagnostic stability, especially in schizophrenia and schizoaffective disorder, even in a FEP sample with a relatively short duration of untreated psychosis