The yield of ictal, postictal, and interictal SPECT was compared in the localisation of seizure foci in 177 patients with partial epilepsy. In 119 patients with known unilateral temporal lobe epilepsy ictal SPECT (97% correct localisation) was superior to postictal SPECT (71% correct), which was better than interictal studies (48% correct). Similarly, in cases of known or suspected extratemporal epilepsy the yield of ictal SPECT studies was high (92%). By contrast, the yield of postictal studies was much lower (46%) and usually only very early postictal studies were diagnostic. Interictal SPECT was of little value. The accuracy of ictal SPECT in localising temporal lobe seizures is now well established. Extratemporal seizures are often brief and difficult to localise. This report shows that ictal SPECT also has a high diagnostic yield in a wide range of extratemporal epilepsies. The brevity of many extratemporal seizures means that true ictal SPECT examinations can be difficult to achieve, but the high diagnostic yield justifies the special organisational effort needed to obtain such studies.