Transcranial magnetic stimulation (TMS) is now accepted as a standard tool for stimulating the human brain through the intact scalp in -elds ranging from preoperative neurosurgical mapping of cortex to cognitive neuroscience or therapy for psychiatric disease. Yet although the basic principles of the method are well understood, we still have relatively little information about which neurons are likely to be activated and how this interacts with ongoing activity in cortical circuits. Much of the uncertainty comes from the fact that the vast majority of experiments have been performed in humans using indirect methods; little detailed work is available in animal preparations. However, since activation by TMS depends on the anatomical arrangement of neural elements in the brain with respect to the position of the coil on the surface of the scalp, it seems likely that even if animal data were available, it could only provide part of the answer to questions about stimulation in the human brain.