Lie detection has been a topic of interest to both scholars and "doers" for millennia, and has acquired more recent urgency given the nature of the "war on terror". Earlier efforts, from the Inquisition's "Dry bread" test to modern polygraphy, capitalized on the autonomic nervous system's homeostatic efforts to compensate for the neuronal upheaval presumably generated by the lie. However, these methods are limited and will not work unless the lie does generate some emotional reaction. Neuroimaging, especially fMRI, makes feasible a new approach whereby thought processes are traced, temporally and spatially, enabling more complex methods of separating lie from truth statements. I will report a series of studies demonstrating unprecedented sensitivity of the fMRI to both lie detection and truth verification, not necessarily perfect opposites. Attentional value of the items used in a deception model, and the ecological relevance of the model to real life deception, can substantially influence the size and directions of the differences between lie and truth observed with fMRI. However, lie can be discriminated from truth on the individual event level, suporting the feasibility of fMRI based practical lie-detection system. However, the query procedure used to elicit deception must be carefully controlled for the confounding effects of attentional salience and the potential effectiveness of counter-measures depends on our ability to gauge this impact.