Background. Recent research has focused upon the subdiagnostic level in an effort to derive more valid domains of psychotic disorder. This has led to the influential positive–negative dichotomy in schizophrenia being superseded by a three-syndrome model. The strategy of looking for syndromes within poorly validated diagnostic categories, such as schizophrenia, has limitations, particularly since it originated in, and has been largely restricted to, the more chronic subsamples.Method. A representative sample of first episode psychosis (N=509), which includes the full spectrum of functional psychosis, was utilized to re-examine the dimensional structure of functional psychosis from first principles. Patients were assessed with the Royal Park Multidiagnostic Instrument for Psychosis (MIP), a comprehensive procedure that documents the psychopathology of the first episode in a clinically valid manner.Results. Principal axis factor analysis was carried out on the tetrachoric correlation matrix of 92 core psychopathological items. A robust and clinically valid four-factor solution was obtained, comprising depression, mania and only two other factors. The first was a Bleulerian blend of negative symptoms, catatonic/motor symptoms and disorganization. The second was a combination of Schneiderian first rank symptoms, and other hallucinations and delusions. The data thus failed to support the three-syndrome model for non-affective symptoms in this population. A six-factor solution, although partially consistent with other studies, represented a more complex and confusing elaboration of the more clinically valid four-factor solution.Conclusions. The findings have implications for the conceptualization of early psychosis, which need to be explored further in validation studies.