Magnetic resonance imaging (MRI) has demonstrated abnormalities of brain structure, particularly of the temporal lobes, in schizophrenia. These are thought to be neurodevelopmental in origin, but when they become evident is unknown.

To determine iftemporal lobe volumes reduce during the development of symptoms of schizophrenia in initially well people at high riskofthis disorder.

A group of 66 people who had at least two first— or second-degree relatives with schizophrenia and a control group of 20 healthy people had a structural MRI scan ofthe whole brain which was repeated after approximately 2 years. Regions of interest, specifically the amygdala-hippocampus complex and the temporal lobes, were traced semi-automatically by three masked raters with good inter— and intrarater reliability

Regional brain volume changes over 2 years did notdiffer between high-risk and healthy participants. Within the high-risk group, the 19 people with psychotic symptoms (12 at first assessment) had a mean reduction of 2163 mm3 intherighttemporal lobe compared with 97 mm3 in the 47 without symptoms (P⩵0.02).

Our findings suggest that people at high risk of schizophrenia with psychotic symptoms show reductions in temporal lobe volumes.

Numerous in vivo brain imaging studies suggest that cerebral Structure is abnormal in schizophrenia, but implicate different regions to varying extents.

We identified published MRI studies in schizophrenia with searches of the computerised literature and key Journals. Reports giving the volumes of cortical structures in people with schizophrenia and controls were included. The percentage differences in volumes were calculated and the median taken as a summary measure for each brain region.

Forty relevant studies were identified. The median percentage volume differences revealed overall reductions in the whole brain (3%), temporal lobe (6% left. 9.5% right), and the amygdala/hippocampal complex (6.5%, 5.5%); and increases in the lateral ventricles (44%, 36%), that were greatest in the body and occipital horns. Segmentation studies suggest that grey matter is reduced but that white matter volumes may actually be increased. In men, substantial reductions were also evident in the amygdala and hippocampus, as well as the largest reductions of all in the parahippocampus (14%, 9%). Few studies gave figures for women alone.

Several brain structures in schizophrenia are affected to a greater extent than expected from overall reductions in brain volume. Further studies are required in affected women, and to try to identify clinical and aetiological associations of these findings.