The lack of robust water markets makes it difficult to value irrigation water. Because water rights are appurtenant to land, it is possible to infer the value of water from observed differences in the market price of land. We use panel data on repeat farmland sales in California's San Joaquin Valley to estimate a hedonic regression equation with parcel fixed effects. This controls for sources of omitted variables bias and allows us to recover the value of irrigation water to landowners in our sample. We show that a more traditional cross-sectional regression results in an artificially low value of irrigation water. © The Author (2014).