The study of electroencephalography (EEG) for deception detection has long been regarded as an alternative to the standard polygraphy, whose main shortcoming is its unacceptably low level of reliability. Most of the EEG deception research has focused on the amplitude and topography of P300. However, the characteristics of the P300 component are tightly connected to the experimental design and hence countermeasures are easily available for P300-based deception detection. The goal of this study is to evaluate different epoching criteria for the extraction of EEG features that are most suitable for the discrimination between truthful and deceptive responses. In order to reach this aim, a modified version of the Guilty Knowledge Test was used where EEG recordings were obtained from four frontal electrodes and two midline electrodes. In none of the electrodes the P300 component differed between deceptive and truthful responses. Differences have instead been revealed through the extraction of response-locked epochs and analysis of area under the curve.