Abstract: A precise extraction of the Visual Evoked Potential (VEP) signal from the Electroencephalogram(EEG) background noise is important to detect human nervous system's disease. Due to the extremely low signal-noise-ratio, however, such kind of signal extraction is by no means an easy tasks. Based on the EEG signal's inherent quasi-periodic feature analysis and statistical evaluation of the related parameters, this paper proposes a speciacl phase-shifting stimulation strategy to significantly enhance the signal-noise-ratios of the superimposed result, which may thus be used as a much better source signal for further processing. Incorporating with other new techniques for extracting the VEP signal to be addressed in the remaining papers by the authors, about ten times of stimulation is enough to obtain a satisfactory result. Some criteria and application examples of the proposed strategy are given in detail.