An anatomical and functional model of drug-resistant partial seizures is presented and discussed based on research conducted by our team over the last decade. This research is based on the study of intracerebral stereoelectroencephalography (SEEG) recordings in an attempt to identify the neural networks involved in generating paroxystic activities so as to understand their dynamics in space and time, and to propose targeted therapies that could "control" these networks. Today, the classical notion of epileptic focus should be replaced by a more complex model that takes into account the potential interactions within the neuronal networks involved in seizures. During partial epileptic seizures, the cerebral structures involved are the seat of characteristic oscillations that may be synchronized or, on the contrary, that can desynchronize in a transitory manner. These epileptic rhythms disturb the physiological rhythms that underlie the cognitive and emotional processes, which can thus be altered in partial epilepsy, even if located far from the original discharge site. We suggest that seizures originate in a group of structures that are highly epileptogenic (epileptogenic zone network, [EZN]) whose activity is synchronized before the appearance of fast oscillations that are transitorily desynchronized. Later, other cortical and subcortical structures are the seat of slower, synchronized rhythmic modifications (propagation network, [PN]). The emergence of clinical signs in the seizure depend on these phenomena, which in some cases can mimic a normal cognitive process or, on the contrary, lead to a deep rupture in normal cerebral functioning.