Hippocampal shape anomaly (HSA), characterised by a rounded hippocampus, has been documented in congenital malformations and epileptic patients. Subtle structural hippocampal abnormalities have been demonstrated in patients with schizophrenia. We tested the hypothesis that HSA is more frequent in schizophrenia, particularly in patients from families multiply affected by schizophrenia, and that HSA is transmitted within these families. We also aimed to define the anatomical features of the hippocampus and other cerebral structures in the HSA spectrum and to determine the prevalence of HSA in a control group. We reviewed the magnetic resonance imaging of a large number of subjects with schizophrenia and bipolar disorder, many of who came from multiply affected families, relatives of the affected probands, and controls. Quantitative measures of hippocampal shape and position and other qualitative anatomical measures were performed (including depth of dominant sulcus cortical cap, angle of dominant sulcus and hippocampal fissure, bulk of collateral white matter, prominence of temporal horn lateral recess and blurring of internal hippocampal architecture) on subjects with HSA. A spectrum of mild, moderate and severe HSA was defined. The prevalence of HSA was, 7.8% for the controls (n=218), 9.3% for all schizophrenic subjects (n=151) and 12.3% for familial schizophrenic subjects (n=57). There was a greater prevalence of moderate or severe forms of HSA in familial schizophrenics than controls. However, there was no increase in the prevalence of HSA in the unaffected first-degree relatives of schizophrenic patients or in patients with familial bipolar disorder. HSA was rarely transmitted in families. HSA was frequently associated with a deep, vertical collateral/occipito-temporal sulcus and a steep hippocampal fissure. Our data raise the possibility that HSA is linked to disturbances of certain neurodevelopmental genes associated with schizophrenia. However, the lack of any increase in prevalence in the unaffected relatives of patients and the lack of clustering within individual pedigrees argues against this developmental anomaly being commonly associated with genetic predisposition to the illness.