Last year the Food and Drug Administration made itself fodder for late-night comedy when it warned the manufacturer of Cheerios against boasting about some of the cereal's health benefits. "We have determined [Cheerios] is promoted for conditions that cause it to be a drug," the FDA said in a letter to General Mills.

By this logic, consumers would need a prescription to buy a box of the oats.

The letter typifies the FDA's longstanding discomfort with health claims made on food labels, particularly those touting the medical benefits of certain diets and ingredients. FDA staff worry the food claims aren't backed by anything close to the science that supports the medical claims made on drug labels. The two, they argue, shouldn't be treated much differently.

As First Lady Michelle Obama is making childhood obesity her signature cause, the Cheerios episode bears remembering. The advice she gets will likely focus on ways to crack down on junk foods, from taxing potato chips to locking away soda. Yet real progress rests with getting food makers to develop diets that don't just taste good, but are good for us. That starts with enabling them to make more claims around the kinds of healthy benefits we want them to pursue.

Food labels today mostly carry assertions about taste, portion size, or packaging--precisely because the FDA doesn't regulate these attributes. Far fewer foods compete on the ability to improve health. When they do, the FDA now often forces them to make the claims so weak and convoluted that the information is unusable for consumers, and marketers.

Yet science is giving us unique opportunities to leverage diet in health. For example, the emerging science of "nutrigenomics" seeks to combine insights from genome research with our understanding of how diet choices affect health. By adjusting food nutrients, the diet choices we make could play a prominent role in mitigating disease by identifying people at risk for certain conditions like cancer and using nutrition as a tool for chemoprevention.

But the FDA could preclude these and other diet innovations by making it too hard to discuss the resulting benefits. The agency has legitimate queries when it comes to translating medical claims onto food labels. But its impulse to apply its standards for drug promotion to all manner of products blurs important distinctions. If a drug touts a partial benefit, then consumers could be misled to forgo more effective therapies. If Cheerios claim heart benefits, the worst is that consumers might reduce their intake of Special K.

The courts recognize the distinction. The FDA has lost three federal court decisions challenging its restrictive labeling policies.

In Pearson v. Shalala (1999), the D.C. Circuit Court of Appeals chastised the agency, stating that it was "skeptical the government could demonstrate with empirical evidence" that health claims made with appropriate disclaimers would "bewilder" consumers, as the FDA claimed. The court drafted labeling language the agency could use to describe the weight of the scientific evidence underlying specific claims.

In response, the FDA in 2003 allowed qualified health claims to appear on food labels so long as manufacturers included a disclaimer that described the reliability of the scientific evidence supporting the association between food and health. The policy established four different levels of evidence, from highly reliable to highly improbable, and graded health claims from A to D. The notion was food makers would invest in developing good evidence linking diet and health, since nobody would want to be forced to disclose their medical claims were sketchy and graded a D.

Unfortunately, the FDA has recently dismantled the 2003 policy on its theory that the disclaimers were confusing. The few new health claims it's allowing on labels are typically supported by forward looking, controlled trials.

But these trials are hard to complete when it comes to food, since you can't control diets the same way you can regulate experiments with a new drug. The FDA has essentially reverted to its restrictive policy prior to the Pearson decision. Some food makers are threatening to sue, again.

For the makers of Cheerios, their mistake was boasting that the cereal could help reduce the risk of heart disease "by lowering the 'bad' cholesterol." That language sprung from accepted research showing that diets high in soluble fiber from whole grain oats can indeed reduce the risk of coronary heart disease. Cheerios' crime was making specific reference to cholesterol levels, something that's typically treated by a drug.

The best course would be for the FDA to allow qualified health claims on food products as it did after 2003. The agency also could go a step further and create a new section on labels to highlight evidence linking food to better health. Consumers would pay more notice to products with health benefits, especially foods that help prevent the causes of and consequence from obesity, as much as the junky foods that didn't have any of these advantages.

As Mrs. Obama tackles childhood obesity, she should enlist food makers in her fight, unleashing them to compete on nutritional benefits and useful innovation in dietary health.

Scott Gottlieb, M.D., is a resident fellow at AEI.