This paper revisits the problem of estimating ex ante cost functions previously studied by Pope and Just (1996) and Moschini (2001). An ex ante cost function that generalizes their ex ante cost functions is introduced, and an econometric procedure for estimating a flexible approximation to it is developed. That generalized cost function is economically relevant not only for the Just and Pope (1996) choice setting but for general producer risk preferences, general stochastic technologies, and general forms of price uncertainty. An econometric strategy for estimating the resulting cost structure that adapts Moschini s (2001) "full-information" approach is developed. And that is followed by an econometric application to US agriculture.