Motions estimation utilizing EEG can develop devices supporting the elderly and disabilities. This paper investigated a possibility of limb motion estimation using Bereitschafts Potential (BP). BP means voltage variation of EEG just before motions. To estimate, neural network was introduced. Three types of processed signal were used as the input. The types of signal processing were FFT, increment and normalization. The best signal processing of them was normalization, and the accuracy rate was 89.0%.