Although farmers in developing countries are generally thought to be risk averse, little is known about the actual form of their risk preferences. In this paper, we use a relatively large field experiment to explore risk preferences related to sweet potato production among a sample of farmers in northern Mozambique. We explicitly test whether preferences follow the constant relative risk aversion (CRRA) utility function and whether farmers follow expected utility theory or rank dependent utility theory in generating their preferences. We find that we can reject the null that farmers’ preferences follow the CRRA utility function in favor of the more flexible power risk aversion preferences. In a mixture model, we find that about three-fourths of farmers in our sample develop risk preferences by rank dependent utility. We also find that by making the common CRRA assumption in our sample, we poorly predict risk preferences among those who are less risk averse.