Summary: The EEG was monitored in 56 patients during implantation ofan internal cardioverter defibrillator. The purpose of this study was todetermine the main EEG frequency ranges that represent ischemic changes duringshort periods of circulatory arrest. The EEG was recorded with a 16-channelcommon reference montage (Cz). After onset of circulatory arrest, the logspectral changes of three-epoch moving averages were calculated relative tothe baseline spectrum. For factor analysis, 17 EEG periods were selected thatshowed changes progressing to an isoelectrical period. Topographic differencesand the time course of quantitative EEG (qEEG) changes were studied in all 56patients. For each patient the EEG period with the longest duration ofcirculatory arrest was chosen. Factor analysis revealed four factors thatrepresented the spectral EEG changes occurring during circulatory arrest andrecovery. The frequency intervals of these factors were 0 to 0.5 Hz, 1.5 to 3Hz, 7.5 to 9.5 Hz, and 15 to 20 Hz for all channels. Only minor topographicdifferences were found in the power of the spectral changes; the sequence ofevents was similar for all electrode positions. The first EEG change aftercirculatory arrest was an initial increase in α power and a decrease inβ power. On average, after approximately 15 seconds α power started todecrease, β power decreased further, δ-1 power started to increase,and δ-2 power started to decrease. After approximately 25 seconds, theδ-1 power increase appeared to plateau or to decrease. A circulatoryarrest longer than approximately 30 seconds resulted in an isoelectrical EEG.After restoration of the circulation, there was a fast transient increase inδ-1 and δ-2 power, followed by a decrease to baseline. α and βpower showed a more gradual increase in power toward baseline and were thelast to restore after 60 to 90 seconds. In general, the spectral changes inthe α and β frequency ranges were most pronounced and consistent. Inconclusion, to detect intraoperative cerebral ischemia, monitoring of changesin the four frequency ranges found is preferable to monitoring changes in theclassically defined frequency bands. Furthermore, these results stress theimportance of the α and β ranges in detecting cerebralischemia.