The main goal of electroencephalography (EEG) based brain-computer interface (BCI) research is to develop a fast and higher classification accuracy (CA) rate method than those of existing ones. Generally, in BCI applications, either motor imagery or event-related P300 based techniques are used for data recording. The stimulus duration (SD) and the inter-stimulus interval (ISI) are crucial two parameters directly affecting the decision speed of the BCI system. In this study, we investigated the performance of the P300 based application in terms of speed and CA for three kinds of protocols which are called fast, medium, and slow included different SD and the ISI values. The training and test data sets were recorded in one week of delay from 8 subjects. The features were extracted by standard deviation, variance, mean, Wavelet Transform and Fourier Transform techniques. Afterwards, they were classified by the k-nearest neighbor algorithm. We obtained 87.08%, 85.41% and 83.95% average CA rate for the fast, medium, and slow protocols, respectively. The obtained results showed that the proposed fast protocol method achieved CA rate between 78.33% and 93.33%. Based on the obtained results, it can be concluded that the fast protocol values can be used for establishing a more accurate and faster P300 based BCI.