To assess the contribution of ictal SPECT to the definition of the epileptogenic zone (EZ) prior to surgery in focal drug-resistant epilepsies, we investigated the effect of the timing of injection and seizure semiology on patterns of perfusion and cerebral blood flow changes (CBF) beyond the EZ. In the rat model of amygdala-kindled seizures, we measured CBF changes with the quantitative [(14)C]-iodoantipyrine autoradiographic method during secondary generalized (SGS, n=26 fully-kindled rats) and focal seizures (FS, n=19 partially kindled rats), according to sequential timing of injection with respect to seizure onset. During SGS, the correct lateralization and rough localization of the focus within limbic structures was only possible at the early ictal and post-ictal times, in between we observed widespread rCBF increases. The switch from hyper to hypoperfusion occurred at the time of late ictal injection. The accurate localization of the EZ was obtained in the study of the more subtle FS (stage 0). At stage 1 of the kindling, there was already a remote widespread spreading of hyperperfusion. In patients surgically cured from a mesio-temporal lobe epilepsy (mean post-operative follow-up: 66 months), we retrospectively studied 26 pairs of ictal and interictal pre-operative SPECTs, classified in 3 groups according to the progression of ictal semiology. Using visual analysis of subtracted SPECTs (SISCOM) and group comparisons with a control group (using SPM), we observed more widespread combined hyper and hypoperfusion with the increasing complexity of seizures. In simple partial seizures, the SISCOM analysis allowed a correct localization of the focus in 4/8 patients, whereas the SPM analysis failed to detect significant changes, due to individual variation, spatial normalization and small magnitude of CBF changes. In complex partial seizures with automatisms, SISCOM and SPM analysis showed antero-mesial temporal hyperperfusion (overlapping the EZ), extending to the insula, basal ganglia, and thalamus in the group of patients having dystonic posturing (DP group) in addition to automatisms. Ictal hypoperfusion involved pre-frontal and parietal regions, the anterior and posterior cingulate gyri, to a greater extent in the DP group. In both human and animals studies, we observed a correlation between the extent of composite patterns of hyper/hypoperfusion and the severity of seizures, and the recruitment of remote sub-cortical structures. Hypoperfused areas belong to neural networks involved in perceptual decision making and motor planning, whose transient disruption could support purposeless actions, i.e. motor automatisms.