Evidence suggests that we use the same mechanisms for both producing and perceiving actions. Such 'shared representations' may also underlie social perception and empathy. However, this idea raises some important and as yet unresolved questions: (i) how do we distinguish other-orientated empathic responses from a self-orientated reactions such as personal distress and (ii) what are the neural substrates underpinning these processes? We employed event-related functional magnetic resonance imaging (fMRI) to explore whether 'shared representations' were recruited to decode dynamic social stimuli in 12 healthy volunteers. We used an adapted version of the Profile of Non-Verbal Sensitivity (Rosenthal, H., Hall, J.A., DiMatteo, M.R., Rogers, P.L., Archer, D., (1979). Sensitivity to nonverbal communication: the PONS test. The Johns Hopkins University Press, Baltimore) which taps social perception using brief silent video clips. Participants chose one of two words that best described the state conveyed by the actor, or in the control condition using the same clips, the word describing which parts of the body were visible (non-social labelling). Off-line self-report measures of empathy and personal distress engendered by thoughts about others, were also given along with an experimentally-derived index of the degree of self-other overlap during social perception. Brain activation specific to the main experimental condition was found in the inferior frontal gyrus (BA44) and premotor areas (BA6) consistent with the use of 'shared representations'. Somatosensory areas such as the insula and supramarginal gyrus (BA40) were also activated suggesting that participants constructed a qualitative representation of the target state. Activity in the rostral anterior cingulate was associated with self-reports of personal distress and increased blood flow to the anterior cingulate (BA24) and inferior parietal cortex (BA40) was related to self-other overlap.