Ictal EEG can be identified by sustained electrographic discharges showing onsets and offsets being clearly distinguished from background rhythm, consisting of rhythmic discharges of variable frequencies and morphology showing gradual variations in frequency and amplitude. Morphology of ictal onset patterns in focal epilepsies are quite diverse and varies even in a same patient. For localizing purposes, onset patterns of ictal EEG are most important than following ictal discharges and they are topographically divided into focal, regional, lateralized, and diffuse. From the electrophysiological point of view, scalp ictal EEG in focal epilepsy requires a source area of synchronized EEG activities spreading over at least 10 cm2, indicating that an ictal onset usually reflects already spreaded activities from the EZ, thus the localization of EZ should not be based upon ictal EEG onset alone but should require its correlation with other investigations. Considering that scalp EEG reflects electrical activities from the cortical surface, ictal discharges arising from the EZ locating in deep sulci or deep cortical structures are not readily picked up by scalp EEG until they spread a significant distance to the cortical surface under the recording electrodes. Another caveat of ictal EEG is that the degree of synchronization is difficult to recognize from the surface EEG. In cases of poor synchrony, scalp EEG may not detect any clear focal or regional discharges but showing widely spread arrhythmic activities, thus fail to carry any localizing information. The lateralizing value of ictal scalp EEG seem to be higher in temporal lobe epilepsy than extratemporal lobe epilepsy, which were reported 76% to 83% and 47% to 65%, respectively. The probability of false lateralization was around 1% to 7% with remaining proportion of ictal EEG being unhelpful due to contamination by severe artifacts or diffuse discharges without focal features. The source localization of ictal onset rhythm by high-density EEG is an emerging technique that seems to perform similarly well to the source localization of interictal spike, which has been proved in large patient cohorts to be accurate and clinically relevant, with positive and negative predictive values rivaling those of structural MRI, although there remain questions regarding the processing of signals for reliable results. A further promising development in seizure onset localization is the inclusion of directed connectivity analysis: thus, not only the most active source is taken into consideration when localizing the onset of a seizure, but also how strongly each source broadcasts its activity to other brain regions (the so-called outgoing connections). Such an approach was shown to be more accurate than merely considering the most active source.