The gole of the research is to consider the information content in application of different fractal methods of deterministic chaos to the automated recognition of sleep phases in computer electroencephalogram (EEG). The Hurst normalized range method, the method of calculating of Grassberger — Procaccia correlation integral and the approximated entropy method are used during the research. The research reveals that a hypnogram can be obtained. It’s possible if appropriate parameters of the methods indicated above are used, as well as the necessary normalization of the original data and averaging the results. A hypnogram has a total coincidence of defned sleep phases for half of the epochs which are recorded on EEG. Current methods of automatic recognition of sleep stages based on the deterministic chaos allow to reduce signifcantly the time of interpretation of polysomnographic recording and reduce the number of channels through which parameters of sleep are registrated.