Reliable and accurate EEG monitoring helps anaesthesiologists to navigate general anaesthesia and to avoid anaesthetic levels that may lead to unwanted awareness or that can cause an increased risk for a postoperative neurocognitive disorder [
1]. In the context of clinical EEG monitoring almost exclusively spectral analytical approaches are applied. The most widely used system, the bispectral index BIS (BIS; Medtronic, Dublin, Ireland) seems to focus on the power in slow and especially in fast frequency bands [
2,
3]. The SEDline (Masimo Corporation, Irvine, CA, USA), Narcotrend (Narcotrend-Group, Hannover, Germany) and Conox (Quantium Medical, Fresenius, Bad Homburg, Germany) monitor also incorporate EEG band power in their algorithm to derive the processed EEG index [
4‐
6]. One monitoring system, the entropy module (GE Healthcare, Helsinki, Finland) applies the Shannon Entropy [
7] to evaluate the shape of the power spectrum [
8]. Since this step occurs in the frequency domain it should not be confused with the time-domain-based permutation entropy (PeEn) [
9], the parameter used in this manuscript. The mentioned monitoring systems evaluate anaesthetic-induced changes in the (power) spectrum of the EEG and calculate an index that inversely correlates with the anaesthetic level or the so-called depth of anaesthesia, albeit this term may not accurately reflect the induced effects of an anaesthetic [
10]. To calculate these indices, the information regarding the (power) spectrum is derived by a linear transformation, the Fourier transformation, to obtain the trigonometric functions, the (stationary) signal can be described with. Although the EEG is nonstationary, episodes up to ~ 25 s may be considered stationary [
11‐
13]. But as electrical activity in the brain exhibits complex behavior with chaotic, non-linear dynamic properties, interpretation of EEG data with non-linear, entropic analyses could be useful [
14,
15]. For scientific purposes, entropy and complexity measures analysing the EEG in the time domain have been successfully applied to perioperatively recorded data. Especially PeEn proved capable in reliably separating EEG from conscious and unconscious states. PeEn codes the time series into order patterns and seems capable of detecting nonlinearities in the signal [
16]. It evaluates the probability distribution of ordinal patterns. Studies showed that various versions of the PeEn perform outstandingly in terms of higher coefficient of determination, prediction probability and less baseline variability compared to other current clinical indices for monitoring the anaesthesia induced with GABAergic agents [
17]. As the calculation of PeEn depends on parameter settings like order pattern length and the number of data points of the EEG segment to be investigated [
17‐
19], we decided to evaluate the impact of order pattern length and the length of the EEG segment on the resulting PeEn. To get a better understanding of PeEn behavior in general, we also investigated the occurrence probability of strictly monotonous patterns as well as non-occurring patterns and how these probabilities change with anaesthesia. We based our analyses on simulated data as well as on EEG recordings from volunteers.