Oil-spill fate and transport modeling may be used to evaluate water column hydrocarbon concentrations, potential exposure to organisms, and impacts of oil spills with and without dispersant use. Important inputs to transport modeling for such analyses are current velocities and turbulent dispersion coefficients. Fluorescein dye studies off San Diego, California, were used to calibrate an oil transport model by hindcasting movement and dispersion of dye. The oil spill model was then used to predict subsurface hydrocarbon concentrations and potential water column impacts if oil were to be dispersed into the water column under similar conditions. Field-collected data included surface currents calculated from high-frequency radar data (HF-Radar), near-surface currents from drifter measurements drogued at several depths (1m, 2m, 4m or 5m), dye concentrations measured by fluorescence, spreading and dye intensity measurements based on aerial photography, and water density profiles from CTD casts. As the dye plume quickly extended throughout an upper mixed layer (7–15m), the horizontal dye movements were better indicated by the drifters drogued to a depth near the middle of that layer than the HF-Radar, which measured surface (∼top 50 cm) currents (including wind drift). Diffusion rates were estimated based on dye spreading measured by aerial photography and fluorescence-depth profiles. The model used these data as inputs, modeling of wind-forced surface water turbulence and drift as a function of wind speed and direction (based on published results of fluid dynamics studies), and Stokes law for droplet rise/sinking rates, to predict oil transport and dispersion rates within the water column. Use of such diffusion rate data in an oil fate model can provide estimates of likely dispersed oil concentrations under similar conditions, which may be used to evaluate potential impacts on water column biota. However, other conditions with different patterns of current shear (due to background currents, tidal currents, and wind stress) should be examined before these results can be generalized.