We present a new three-dimensional (3-D) P-wave velocity model for the upper-crustal struc-ture beneath the Strait of Georgia, southwestern British Columbia based on non-linear tomo-graphic inversion of wide-angle seismic refraction data. Our study, part of the Georgia Basin Geohazards Initiative (GBGI) is primarily aimed at mapping the depth of the Cenozoic sed-imentary basin and delineating the near-surface crustal faults associated with recent seismic activities (e.g. M = 4.6 in 1997 and M = 5.0 in 1975) in the region. Joint inversion of first-arrival traveltimes from the 1998 Seismic Hazards Investigation in Puget Sound (SHIPS) and the 2002 Georgia Basin experiment provides a high-resolution velocity model of the subsurface to a depth of ∼7 km. In the southcentral Georgia Basin, sedimentary rocks of the Cretaceous Nanaimo Group and early Tertiary rocks have seismic velocities between 3.0 and 5.5 km s −1 . The basin thickness increases from north to south with a maximum thickness of 7 (±1) km (depth to velocities of 5.5 km s −1) at the southeast end of the strait. The underlying basement rocks, probably representing the Wrangellia terrane, have velocities of 5.5–6.5 km s −1 with considerable lateral variation. Our tomographic model reveals that the Strait of Georgia is un-derlain by a fault-bounded block within the central Georgia Basin. It also shows a correlation between microearthquakes and areas of rapid change in basin thickness. The 1997/1975 earth-quakes are located near a northeast-trending hinge line where the thicknesses of sedimentary rocks increase rapidly to the southeast. Given its association with instrumentally recorded, moderate sized earthquakes, we infer that the hinge region is cored by an active fault that we informally name the Gabriola Island fault. A northwest-trending, southwest dipping velocity discontinuity along the eastern side of Vancouver Island correlates spatially with the surface expression of the Outer Island fault. The Outer Island fault as mapped in our seismic tomog-raphy model is a thrust fault that projects directly into the Lummi Island fault, suggesting that they are related structures forming a fault system that is continuous for nearly 90 km. Together, these inferred thrust faults may account for at least a portion of the basement uplift at the San Juan Islands.