Lie detection has recently become a topic of discussion once more. Courts of law were interested in it for a long time, but the unreliability of the polygraph prevented any serious use of it. Now a new technology of mind-reading has been developed, using different devices that are deemed to be able to detect deception, in particular Functional Magnetic Resonance Imaging (fMRI). Is fMRI more reliable than the polygraph? It meets at least with various kinds of obstacles: technical, methodological, conceptual and legal. Technical obstacles are linked with the state of the technique, methodological ones with epistemological difficulties, conceptual ones with problems tied to what lying consists of, and legal ones with the effects of brain imaging on lawsuits. I examine several of these and conclude that at present mind-reading using fMRI is not ready for use in the courts. The obstacles examined may not be insuperable, but a lot more research is needed.