When Eve Johnstone and colleagues (1976), and later Weinberger et al. (1979a,b) reported that patients with schizophrenia showed evidence of morphological brain abnormality, many interested researchers expressed considerable scepticism (Gelenburg, 1976; Jellinek, 1976; Marsden, 1976). Were the abnormalities attributable to medication, institutionalization, ageing or simply an epiphenomenon?

There is now considerable evidence that schizophrenia is a brain disease characterized by abnormalities in cerebral structure and function (Waddington, 1993; Carpenter and Buchanan, 1994; Weinberger, 1995; Harrison, 1999). Studies using neuroimaging, neuropathology and neurophysiological approaches have now localized these changes to the frontotemporal regions of the brain (Bogerts et al., 1990; DeLisi et al., 1991; Degreef et al., 1992; Bilder et al., 1992; Hoff et al., 1992; Sweeney et al., 1992; Saykin et al., 1994; Salisbury et al., 1998). But why did it take over a decade to lay the sceptics' doubts to rest? Because of sampling: schizophrenia has a high prevalence but low incidence, so most research studies are either cross-sectional in design (mixing incident and prevalent cases) or focus on individuals who have been ill and receiving treatment for many years (Lieberman et al., 1992). The so-called ‘convenience sample’ favours those with established chronic illness at the expense of patients with good outcome psychotic illness who do not remain in contact with services. Therefore, the powerful effects of medication, hospitalization and the chronic schizophrenic process itself are further concentrated in typical research samples.