Besides visual sleep stage scoring, it is possible to separate sleep stages by analyzing spatially distributed three-channel electroencephalogram (EEG) root mean square (RMS) amplitudes. This approach allows qualitative and quantitative evaluation of sleep architecture. The aim of this study was to analyze the effect of apnea–hypopnea index (AHI) and state of vigilance on the microstructure of cortical activity during sleep using EEG cluster analysis. In 31 obstructive sleep apnea (OSA) patients, cortical EEG patterns were recorded during polysomnography and subjected to EEG cluster analysis. The results were subsequently correlated to AHI as well as to subjective (Epworth Sleepiness Scale) and objective (pupillographic sleepiness test) daytime sleepiness using linear regression. In 18/31 patients, cortical EEG patterns were recorded again the following night during continuous positive airway pressure (CPAP) titration and compared to the results of the first night. Linear regression analyses revealed dependencies of specific cortical EEG patterns on AHI (intra-cluster distance of N2 [p = 0.019], N3 [p = 0.024]; inter-cluster distance of wake-N2 [p = 0.046], wake-N3 [p = 0.047], N2–N3 [p = 0.021]) and on objective daytime sleepiness (wake intra-cluster distance [p = 0.047], inter-cluster distance of wake-N3 [p = 0.042]), but not on subjective daytime sleepiness. Specifically, an increase in EEG pattern variability was found in patients with high AHI values, which could be reduced by CPAP therapy. This new approach enables objective analysis and visualization of sleep macro- and microstructure. The variance of neuronal EEG signal patterns enables conclusions on the presence of pathological respiratory events and objectively measured daytime sleepiness.