Although the U.S. spends billions of dollars a year controlling water pollution, there is little empirical evidence of comparable benefits. This study argues that measurement error in national pollution data has caused benefits to be underestimated. Using an instrumental variables approach, the study shows that there are substantial benefits from reducing nutrient pollution. Instrumental variable estimates of the effects of phosphorus on recreational use are an order of magnitude larger than conventional estimates. The study uses a carefully measured state-level pollution dataset to show that this difference is consistent with an estimate of measurement error in national water pollution data.