Abstract
SUMMARY: Two obstacles to achieving diets which maximize farm profits are discussed: (1) the continuing inability to predict reliably the response of animals to a diet with any given composition, and (2) batch-to-batch variation in the true nutrient content of a diet with a fixed nominalcontent. With the aid of simple practical examples it is demonstrated that diet recommendations can be more sensitive than might be expected to the type of function used to predict response, and that variation in dietary nutrient content may reduce farm profits (though the size of the reduction is not clear because of (1) above).