This chapter presents the methods of analysis required for the "oldest" types of brain–machine interface. These methods are (strongly) invasive, as they require trepanation to insert a large number (10–100) of electrodes into the brain tissue. The first feasibility studies were performed on rats and on monkeys . The advantage of these methods of recording is that they give access to individual neuron activity with excellent resolution in time – achieving this resolution is the main subject of this chapter – and the obvious disadvantage is that they require trepanation, which excludes them from being used with patients, except in very exceptional cases. As the other chapters of this book will discuss, BCIs may be implemented without recourse to the invasive methods that we shall discuss here; however, these methods are still very frequently used by neurophysiologists in a wider context that we shall introduce in the following section.