An outline is given of some of the methodological issues discussed in neuroradiological research on psychiatric illness. Strengths and shortcomings of magnetic resonance imaging (MRI) in depicting and quantifying brain structures are described. Temporal lobe anatomy and pathology are easily accessible to MRI, whereas limits on anatomical delineation hamper approaches to frontal lobe study. White matter hyperintense lesions are sensitively depicted by MRI, but specificity is limited. Distinction of vascular and primary degenerative dementia is considerably improved by CT and MRI analysis. Computed tomography (CT) and MRI have enhanced the understanding of treatable organic psychiatric disorders, e.g., normal pressure hydrocephalus. Subcortical and white matter pathology has been replicated in CT and MRI studies of late-onset psychiatric disorders, clinical overlap with cerebrovascular disease or neurodegeneration may be of import. Transcranial sonography findings of brainstem structural change specific to unipolar depression may contribute to the understanding of affective psychoses. Magnetic resonance spectroscopy and functional MRI are likely to stimulate psychiatric research in the future.