Based on detailed aftershock monitoring, the first model of the Athens earthquake is formulated, which is consistent with global, regional, and local strong-motion data of the mainshock, and fits with geological setting. The 30-station temporary network located 450 aftershocks. During the first 20 observation days the aftershocks identified the mainshock fault plane dipping 52° and striking 117°, consistently with the teleseismic fault-plane solution. A formal upward continuation of the fault plane intersects the surface close to the Fili fault. Numerical modeling of the broadband regional data at 10 stations (epicentral distances 140 to 370 km) estimates the centroidal source depth of 10 km and yields an average source duration of 5 to 6 sec. The interstation variability of the apparent duration indicates source directivity. The empirical Green's function modeling at the closest broadband station suggests a fault length of 20 or 10 km. Both the numerical and empirical modelings give a very short rise time of 0.1 to 0.3 sec. The short rise time seems to favor the nearly complete stress release of an asperity. A 10 km asperity (stress drop of 2.7 M Pa) is in agreement with a gap, identified during the first 12 observation days between two aftershock clusters. The strong-motion accelerograms in Athens also indicate a short apparent duration due to directivity (about 3 sec), and confirm an abrupt rupture beginning. There is no evidence for an abrupt stopping. The short rise time and short apparent source duration were two principal factors determining the damaging ground motions in Athens.