The P-300 (commonly abbreviated as P3) is one of the most researched phenomena in the field of cognitive neuroscience. One of the many applications that derived from the discovery of this brain signal is deception detection. This practice commonly utilises the P3 in Concealed Information Tests (also called Guilty Knowledge Tests). As the name suggests, this type of test allows experimenters to detect information that is being concealed by a suspect. An example of such a situation, in practice, could involve a terrorist concealing their identity by pretending to have a different name. This example is particularly relevant to this thesis, which focuses on identity deception. More in detail, we propose the development of a deception detector which utilises Rapid Serial Visual Presentation (RSVP) as a presentation technique. In this way, we present stimuli on the fringe of awareness, greatly reducing the possibilities of applying countermeasures, whose use is an issue in many deception detector implementations. U sing the P3 is advantageous as it can be easily measured by electroencephalography; the high temporal resolution and reduced cost of EEG are appropriate for rapid deployment in field situations. Recorded data can then be analysed utilising Randomisation, a form of Monte Carlo Resampling, particularly useful in performing statistical inferences at the individual level. We tested this implementation and replicated it various times: different types of concealed information were utilised (namely birthdays and first names). As will be demonstrated, our system was successful, achieving high hit rates and low false alarm rates. We also instructed participants to deliberately attempt to confound the deception detector and we explain how our deception detector can resist their application. Finally, we discuss a number of alternative analyses on the presented data. We studied the relationship between the two types ofP3 that were elicited in our experiments: the P3a and the P3b. The randomisation method proposed is also examined under balanced and unbalanced situations. Finally, we compare "standard" ERP analyses with a novel tool that we call Spectral Mass, that employs time-frequency transforms in place of averaging.