Clinical practice, training and evaluation of treatment in the functional psychoses continues to be carried out mostly along the traditional line of separation by diagnostic entity. However, the combined evidence from research on risk factors for onset and for persistence of psychotic illness indicates quantitative, but not qualitative, differences between categories of schizophrenia and affective psychosis. "Developmental" factors, such as childhood dysfunction, increased cerebral ventricle size and familial morbid risk of schizophrenia operate preferentially, though not specifically, at that end of the psychopathological spectrum characterised by a preponderance of negative features. On the other hand, "social" factors, such as ethnic group, adverse life events and familial morbid risk of affective disorder have a larger impact at the end associated with predominance of affective features. Heterogeneity in the functional psychoses may thus be best conceived as two discrete effects operating at different ends of a continuous psychopathological spectrum. The use of highly reliable but arbitrary diagnostic categories may introduce serious bias in aetiological and treatment research. Evidence supporting the validity of a model of shared risk factors for continuous characteristics needs to be further elaborated and incorporated into our concepts of psychotic illness.