The importance of risk in a farmer's decision to use herbicides should not be forgotten. This paper presents a theoretical model of the weed control decision and develops a MOTAD programming model. The model uses herbicides at levels of risk aversion found in previous empirical studies. This result was not changed by equal (or even higher) ASCS yields for not using herbicides, a tax on herbicides, or the elimination of deficiency payments. The variability of returns was more important in influencing a farmer to choose herbicides than the higher expected returns in a nonherbicide system.