We show that the weather premium, an anecdotal phenomenon in the U.S. corn futures market, can arise from a convex demand function. We further show that the magnitude of the weather premium depends on the carryout and expected yield at harvest. We use data from 1968 to 2015 to evaluate the accuracy of the December futures price as a forecast of the harvest price. A predictable component in the forecast error is consistent with the existence of a time‐varying weather premium. We demonstrate that a passive strategy of routinely shorting the corn December futures does not provide an attractive risk‐adjusted return.