This paper presents a review of finite fault modeling, i.e. inversion of the earthquake source mechanism using near-source strong motion records as input. The first earthquakes for which there was adequate strong motion data for such studies were the 1966 Parkfield and 1971 San Fernando, California earthquakes. Since then, the finite fault modeling has become an integral part of the modem earthquake studies. Knowledge of the features of the earthquake source mechanism, as retrieved from near-field strong motion records, is valuable for understanding the physical processeS related to brittle release of tectonic strain during earthquakes, and for prediction of strong ground motion from future earthquakes in seismic design of structures. The paper includes a review of (I) the published literature on this topic, (2) the methodology (theoretical source models and solution of the inverse problem) with emphasis on !be difficulties and limitations, and (3) the strong motion-data that has been used as input in these studies. The paper also comments on the source mechanism of submarine earthquakes, slides and slumps, and discusses the possibility of using maregrams for inversion of their source mechanism.