A random utility maximization (RUM) model is presented for the purpose of providing benefits estimates to aid in the implementation of the total maximum daily load (TMDL) process. The specific application is to North Carolina, for which the 1994 Environmental Protection Agency (EPA) National Demand for Water-Based Recreation Survey is used to characterize the demand for water recreation in the state as a function of ambient water quality. The RUM choice set is defined to include all 8-digit hydrological units or watersheds in the state, allowing benefits estimates to match the spatial aggregate of interest for the TMDL process. Water quality measures are linked to the choice set via the use of EPA's Index of Watershed Index database. A significant relationship between watershed-level water quality and recreation trip taking behavior is found. Per trip and aggregate benefits estimates are presented for watershed, river basin, and statewide improvements in water quality.