Using a dense seismic array in Long Beach, California, we have investigated the effectiveness of using traffic noise for passive subsurface imaging. Spectral analysis revealed that traffic-induced vibrations dominate the ambient seismic noise field at frequencies between 3 and 15 Hz. Using the ambient-noise crosscorrelation technique at these frequencies, we have extracted fundamental- and first-order-mode Rayleigh waves generated by Interstate 405 and local roads. We picked group traveltimes associated with the fundamental mode and used them in a straight-ray tomography procedure to produce group velocity maps at 3.0 and 3.5 Hz. The velocity trends in our results corresponded to shallow depths and coincided well with lithologies outlined in a geologic map of the survey area. The most prominent features resolved in our velocity maps were the low velocities to the north corresponding to less-consolidated materials, high velocities to the south corresponding to more-consolidated materials, a low-velocity zone corresponding to artificial fill in Alamitos Bay, and a low-velocity linear feature in the Newport-Inglewood Fault Zone. Our resulting near-surface velocities can be useful for identifying regions that are susceptible to serious damage during earthquake-related shaking.