MRI has been applied to the investigation of epilepsy for 12 years. The principle role of MRI is in the definition of structural abnormalities that underly seizure disorders. Hippocampal sclerosis may be reliably identified, quantitative studies are useful for research and, in equivocal cases, for clinical purposes. A range of malformations of cortical development (MCD) may be determined. In patients with refractory partial seizures who are candidates for surgical treatment, a relevant abnormality is identifiable using MRI in 85%, it is likely that subtle MCD or gliosis accounts for the majority of the remainder. The proportion of cryptogenic cases will decrease with improvements in MRI hardware, signal acquisition techniques and post-processing methodologies. Functional MRI is used to identify the cerebral areas that are responsible for specific cognitive processes, and is of importance in planning resections close to eloquent cortical areas. Magnetic resonance spectroscopy (MRS) provides a means of investigating cerebral metabolites and some neurotransmitters, non-invasively. The concentrations of N-acetyl-aspartate (NAA), creatine and choline-containing compounds may be estimated using proton MRS. Reduction of the ratio of NAA/(creatine+choline) is a feature of cerebral regions that include epileptic foci. Cerebral concentrations of GABA and glutamate, and the effects of antiepileptic drugs on these, may be estimated. Concentrations of high energy phosphate compounds, inorganic phosphate and pH may be assessed using 31P-MRS. In general, epileptic foci are associated with an increase in pH, increased inorganic phosphate and decreased phosphate monoesters. Carbon-13 spectroscopy promises to be a useful method for investigating cerebral metabolism in vivo. PET may provide data on regional cerebral blood flow (rCBF), glucose metabolism and the binding of specific ligands to receptors. Correlation of functional and structural imaging data is necessary for adequate interpretation. The hallmark of an epileptic focus is an area of reduced glucose metabolism, identified using [18F]fluorodeoxyglucose (18FDG), that is commonly more extensive than the underlying anatomical abnormality. The clinical role of 18FDG-PET requires re-evaluation in the light of the advances in structural imaging with MRI. Specific ligands are used to investigate specific receptors. Benzodiazepine and opioid receptors have been studied most. Reduced benzodiazepine receptor binding is commonly seen at an epileptic focus, in a more restricted distribution than an area of hypometabolism. Focal increases and decreases in benzodiazepine receptor binding have been demonstrated in MCD in areas that appear normal on MRI, indicating the widespread nature of the abnormalities. It has been found that mu-opioid receptors are increased in temporal neocortex overlying mesial temporal epileptic foci. Dynamic studies of ligand-receptor binding are possible using PET, for example the release of cerebral endogenous opioids has been implied at the time of serial absences. The main use of single photon emission computed tomography (SPECT) is to produce images reflecting rCBF. Interictal studies alone are not reliable. A strength of SPECT is the ability to obtain images related to rCBF at the time of seizures. Concomitant video-EEG recording is necessary. Ictal scans need to be considered in comparison with an interictal scan and an MRI. Interpretation must be cautious, but may yield data that is useful in the investigation of patients for possible surgical treatment.