The last several decades have witnessed a rapid increase of people suffering from Alzheimer's disease (AD), one of the most popular causes of disability in late-life. More and more frequently, electroencephalography (EEG) is used to investigate the changes in the brain's information processing in the course of AD in order to find supplements for the clinical diagnostics. In this thesis, the EEG signal is considered as a sequence of electric potential landscapes with stable topography. These so-called microstates are analyzed with respect to their duration, topography, ratio and occurrences. The analysis is performed for 96 AD patients of the PRODEM-AUSTRIA database. Thereby, the severity of AD is measured by using the Mini-Mental-State-Examination (MMSE) score. Within this thesis, two approaches for the determination of EEG microstates are presented. The first strategy determines the centroids of the positive and negative potential areas to distinguish different microstates and thereby segment the EEG sequence. The other approach uses a modified K-means algorithm to cluster all occurring microstates into a predetermined number of classes. The statistical evaluation also consists of two independent parts. In the former, the data are analyzed versus MMSE score by using a least squares regression. In the other, the data of 79 patients are evaluated versus MMSE score by using a least squares regression including demographic variables as sex, age, duration of AD and degree of education. For both statistical approaches and the segmentation as well as the clustering procedure, the results could not document a significant change in the characteristics of EEG microstates in the course of AD. For future research, the impact of longitudinal studies is interesting since the basic characteristics of EEG microstates of AD patients vary a lot. The comparing of past and more actual EEG recordings in relation to the worsening of the AD symptoms and the more severe impairment of one patient could therefore be revealing.