In order to understand what may have triggered the 1995 Kobe, Japan, earthquake and how the rupture proceeded after initiation, we determined high-resolution three-dimensional (3-D) P and S wave velocity and Poisson's ratio structures in the Kobe epicentral area and relocated the aftershocks with the obtained 3-D velocity model. We used 64,337 P and 49,200 S wave high-quality arrival times from 3634 Kobe aftershocks and local microearthquakes recorded by both permanent seismic networks and portable stations that were set up following the Kobe mainshock. Significant velocity variations of up to 6% are revealed in the aftershock area. We found that areas with high aftershock activity are generally associated with low Poisson's ratio, which may be the strong and competent parts of the fault zone and were apt to generate aftershocks. The Kobe mainshock hypocenter is located in a distinctive zone characterized by low P and S wave velocities and a high Poisson's ratio. This anomaly exists in the depth range of 16 to 21 km and extends 15 to 20 km laterally. This anomaly may be due to a fluid-filled, fractured rock matrix that contributed to the initiation of the Kobe earthquake. Our interpretation has been supported by many pieces of evidence from hydrological, geochemical, seismological, and geophysical investigations conducted at the Kobe earthquake region.