Vehicle accidents are rapidly increasing in many countries. Among many other factors, drowsiness is playing a major role in these accidents and systems which can monitor it are currently being developed. Among them, Electroencephalography (EEG) proved to be very reliable. Indeed, many EEG based drowsiness detection techniques are proposed for drivers. Most of these drowsiness detection techniques are normally subdivided into feature extraction and classification methods. Features obtained from FFT are effective and give higher accuracy; but are limited by the non stationary behavior of EEG signals. This paper reviews some of the most recent work of the EEG based drowsiness detection techniques. It shows a major gap found in all these studies, which is the fact that the channel selection method is not clearly specified. Therefore, research can be undertaken to properly choose suitable channel(s) to realize accurate detection of drowsiness. This survey also highlights the fact that, there is no publicly available data and comparison between techniques is not yet possible, because each technique is tested on its own dataset. © 2006-2018 Asian Research Publishing Network (ARPN).