The invention relates to an electroencephalogram chaos characteristic analysis method. The method includes the following steps that: electroencephalogram signals are filtered, and the filtered electroencephalogram signals are decomposed into a plurality of sub-band signals; extreme values points and time points corresponding to the extreme values points of the sub-band signals are extracted, and a monotonic amplitude sequence and a monotonic period sequence are generated according to the extreme values points and the time points; the monotonic amplitude sequence and the monotonic period sequence constitute a vector sequence, and artifact removal is performed on the vector sequence; the vector sequence is divided into a plurality of subintervals in the dimensionalities of monotonic amplitude and monotonic period, so that the probability of each vector distributing in each subinterval can be obtained, and the vibrational entropy of the electroencephalogram signals can be obtained according to the probability; and the chaos degree of the electroencephalogram signals is analyzed according to the vibrational entropy of the electroencephalogram signals. The electroencephalogram chaos characteristic analysis method of the invention can effectively reflect the distribution characteristics of the waveform vibration characteristics of the electroencephalogram signals.