The earthquake of 7 Sept. 1999 (Ms= 5.9) that struck the metropolitan area of Athens, occurred only at a distance of '18 NW from the historical center of the city and has been the most destructive shock in modern history of Greece. Therefore, it is of great importance to identify the seismogenic structure. Focal mechanisms of the main shock as well as the lateral and vertical distributions of the aftershocks , determined by several institutes, are consistent in that the main shock was associated with a normal faulting striking WNE-ESE and dipping SW. The only geological faults known in the area to have the geometrical characteristics that are compatible with the seismographic results are the fault of Thriassion Pedion and the Fili fault. We support that the Fili fault, striking 110° - 150° Ν at an observable length of 8-10 Km, has been very probably the seismogenic structure of the Athens earthquake because ( i) it proved to be an active fault as revealed by the striations we observed on several segments of the fault mirror, ( ii) the meizoseismal region as well as the most important ground failures, like local small-scale landslides and rock-falls, all are located on the hanging-wall domain and very close to the surface trace of the fault as it is theoretically expected (e.g. Oglesby et al., 2000), ( iii) we observed a possibly co-seismic displacement by 3-6 cm of the SW (hanging-wall) segment towards SW . On the contrary, the fault of Thriassion Pedion is recognizable in satellite images but in the field it is evident only as a series of alluvium cones which is an evidence of a possibly inactive structure. Moreover, the meizoseismal area and the ground failures observed in association with the Athens earthquake are located in its foot-wall, that is in the domain where strong motion should not expected to occur. In addition, if that fault was the seismogenic one then the earthquake focus would fit the fault geometry only if it was shifted at least 15 km southwestwards. The last possibility that the Athens earthquake was associated with a blind fault is not supported by any kind of instrumental or field evidence.