A case-control study was undertaken of volumetric computerized tomographic scan measures in 216 consecutive admissions for functional psychosis and 67 healthy community controls. Odds ratio analysis demonstrated significant linear trends in the association between increasing lateral and third ventricle volumes, and both RDC schizophrenia (N = 121) and schizo-affective disorder (N = 41); cases were consistently associated with larger volumes than controls. There was an association between larger third, but not lateral, ventricle size in affective psychoses (N = 54). These associations were statistically independent of intracranial volume, sex, social class and ethnicity, factors which were significantly associated with ventricular measures in the controls. There was no evidence of a threshold corresponding to the notion of normal versus enlarged ventricles. Within the schizophrenia group, there were no large or significant associations between ventricle dimensions and age at onset, duration of illness or pre-morbid social functioning. Neither obstetric complications nor a family history of schizophrenia or other psychiatric illness was associated with large ventricles in these cases.