The effects of lithology, fracturing, and gouge zone mineralization on the geophysical properties of fault zones are not very well understood. In situ seismic data collected over the exhumed San Gregorio Fault at Moss Beach, CA were used to relate in situ compressional wave velocity to internal fault zone properties. This active strike-slip fault is exposed in cross section on an uplifting and actively eroding wave-cut platform. It cuts shallow marine sediments that have been buried to depths of a few kilometers. The unweathered exposure containing seawater makes it a unique analog of subsurface faults. Previous structural analysis over this exposure observed damage caused by faulting over a ~100 m wide zone in cross-section. The fault zone is centered at a 10-17 m wide clay-rich fault core flanked by a ~30 m wide brecciated gouge zone. These gouge zones are bordered on either side by 30-40 m wide fractured zones. Resolving to a scale of a few meters, the seismic survey produced a continuous P-wave velocity profile analogous to a horizontal well log across the fault. Lateral variations in the velocity profile correlate exactly to previously mapped fault zone structure. The clay core and adjacent brecciated gouge create a ~50 m wide very low velocity zone, 25-50% slower than the surrounding host rock. Fractured bedrock on either side of the core causes a wider zone of 5-10% slow velocity, for a total fault signature ~100 m wide. Fault parallel fracture anisotropy was observed in the fractured zones, but surprizingly anisotropy was not observed in the strongly foliated gouge zones. The field measurements differ significantly from laboratory measurements at zero pressure and in some cases from expected values for saturated rock of this porosity, perhaps due to biased rock sampling, the long wavelength effects of macrofractures, frequency dispersion, and partial saturation. The velocity profile is similar in width and consistent in velocity contrast to low S-wave velocity zones derived from fault zone guided waves in other strike-slip faults. The traveltime delay across the fault zone is not large enough to cause the 2-3 km wide crustal low velocity zones modeled by refraction studies. Synthetic reflection seismograms in the typical frequency range show that the fault zone acts as a thick bed or as a constructively interfering thin bed. The models suggest that very large reflection coefficients observed across accretionary prism faults can be explained by fracturing, brecciation and clay content without elevated pore pressures. Comparison with a refraction study across the Punchbowl Fault shows a similar structural zonation of these two well-studied examples of brittle fault zones. This suggests that high-resolution seismic velocity models can be used to directly interpret internal deformation structure of brittle faults.