Leonard Berlin reports that neuroscientific data play an increasing role in court. They have been used to argue that criminals are not morally responsible for their behaviour because their brains are 'faulty', and there is evidence that such data lead judges to pass more lenient sentences. I raise two concerns about the view that neuroscience can show criminals not to be morally responsible: That the brains of (say) violent criminals differ from most people's brains does not straightforwardly show that violent criminals are less morally responsible. Behavioral states arise inter alia from brain states, and since violent criminals' behavioral states differ from those of most people, it is unsurprising that violent criminals' brains should differ from most people's brains. This no more shows violent criminals to have diminished moral responsibility than differences between the brains of cheerful and uncheerful people show either group to have diminished moral responsibility.Those who view brain abnormalities as evidence of reduced moral responsibility rely on the assumptions that people with normal brains have free will and that we know what sorts of brain activity undermine free will. However, both of these assumptions are highly controversial. As a result, neuroscience is not a reliable source of information about moral responsibility. I conclude that, until we settle whether and under what circumstances brain activity is incompatible with free will, neuroscience cannot tell us anything useful about criminal accountability.