The GI has been around since 1981, when nutrition scientist David Jenkins, MD, PhD, set out to determine which carbs are best and which are worst. First, he had to come up with a way to measure a food’s effect on blood sugar.

In a study, published in The American Journal of Clinical Nutrition, Dr. Jenkins and a team of researchers enlisted a group of healthy volunteers and asked them to eat a variety of common foods, each of which contained 50 grams of carbohydrate. The researchers then measured the participants’ blood sugar over the two hours that followed.

The results were surprising. Almost everyone assumed that table sugar would be the worst offender, certainly worse than the complex carbohydrates found in starchy staples such as rice and bread. But this didn’t always prove true. Some starchy foods, like potatoes and cornflakes, ranked very high on the index, raising blood sugar nearly as much as pure glucose. Where the glycemic index fell short?

The results seemed to demonize healthy foods, such as carrots and strawberries. Watermelon was practically off the top of the GI chart. But no one ever gained weight from eating carrots. Nor do carrots, in the real world, raise blood sugar. What was the GI missing?

The GI measures the effects of a standard amount of carbohydrate: 50 grams But you’d have to eat seven or eight large carrots to get 50 grams of carbohydrate. The same holds true for most vegetables and fruits. They’re full of water, so there’s not much room in them for carbohydrates. Bread, on the other hand, is crammed with carbohydrates. One large slice has 48 grams of total carb.

That wasn’t the only problem. “The glycemic index rating applies only when the food is consumed on an empty stomach” without any other type of food, says Sheth. That isn’t exactly the way most people eat. And it explains why the American Academy of Nutrition and Dietetics considers the glycemic index an imperfect, though useful tool in identifying lower-glycemic foods.