A comparison of accepted nutritional advice with actual American dietary practice suggests that many people fail to eat well in spite of well-documented health consequences. Popular culture often labels the worst offenders as lacking in “self-control,” and many blame the aggressive advertising campaigns of the fast-food and snack-food industries for manipulating consumers into poor diets, but these conclusions are not easily reconciled with a neoclassical approach to economic decision theory. This essay considers the consumer’s “diet problem” in light of emerging evidence from a number of behavioral sciences. In particular, it is argued that human evolution in the distant past resulted in an elegant solution to this problem (of search for a suitable diet in an uncertain environment), which any neoclassical economist would recognize. In modern environments, however, the signals that formerly provided information in the consumer’s search problem are subject to manipulation by food-producing firms. Confirmation by molecular biologists that many human responses to these signals are firmly encoded in our genes suggests a need to re-evaluate the welfare economics of the food industry.