Using computed tomography, lateral ventricular size was studied in a sample of 112 institutionalized chronic schizophrenic patients (selected from 510 cases to investigate the correlates of the defect state and intellectual decline and the effects of insulin, electroconvulsive and neuroleptic treatment), and compared with matched groups of non-institutionalized schizophrenics, patients with first schizophrenic episodes, institutionalized and non-institutionalized patients with primary affective disorder, and neurotic out-patients. Age was significantly correlated (P less than 0.0002) with lateral ventricular size, but the institutionalized schizophrenic patients had significantly larger (P less than 0.025) lateral ventricles than the neurotics when age was taken into account. Ventricular enlargement was unrelated to past physical treatment (neuroleptics, insulin coma and electroconvulsive therapy). Within the group of institutionalized schizophrenic patients few correlates of ventricular enlargement were identified; thus in this population increased ventricular size was not clearly associated with the features of the defect state (negative symptoms and intellectual impairment). However, there was a curvilinear (inverted-U) relationship between intellectual function and ventricular size. Increased ventricular size was significantly related to absence of hallucinations, impairment of social behaviour, inactivity and the presence of abnormal involuntary movements. The findings confirm that structural brain changes do occur in chronic schizophrenia, but illustrate some of the difficulties in elucidating the clinical significance of ventricular enlargement. Lateral ventricular size is strongly age-related and the distribution in chronic schizophrenia is skewed and not bimodal; the relationship to particular features of the disease is complex and likely to emerge only in studies with a large sample size.