Abstract : System identification consists in building mathematical models of dynamical systems from experimental data. Such a methodology was mainly developed for designing model-based control systems. More generally, parameter estimation is at the heart of many signal processing applications aiming to extract information from signals, like radar, sonar, seismic, speech, communication, or biomedical (EEG, ECG, EMG) signals. Nowadays, dynamical models and identification methods play an important role in most of scientific disciplines such as automatic control, signal processing, physics, economics, medicine, biology, ecology, seismology, etc. In this plenary talk, an overview of the principal models and identification methods will be presented.