This research explores the effectiveness of different combinations of preprocessing and feature extraction technics in identifying mental activities representing right/left hand movement using Electroencephalography (EEG) data. The EEG signals used were recorder for a subject during left and right hand movement through 19 channels (electrodes) at 500Hz. However, only channel C4 and C3 were used in the classification. A three-layered feedforward artificial neural network, which implements the backpropagation of error learning algorithm, was used to preform the classification in the four methods that were tested. The first one obtained a success rate of 68.4% by feeding the signals resulted from separating Mu brainwaves using FIR filter. Preprocessing in the rest of the methods was done by employing the FIR filter to clean the EEG from deferent kind of artifacts. The absolute values of the Fourier transform of the EEG were used as the input to the classifier in the second method resulting in an accuracy of 81.6%. Whereas in the third method, the classifier used the energy, variance, and waveform of the wavelet decomposition for ten levels to achieve a 63.2% success rate. Finally, the forth method revealed the superior success rate of 94.7% by classifying the coefficient of the ten levels decomposition.