I’ve had enough questions about the glycemic index and the glycemic load that I’ve decided to take the time and explain what it all means.  If you know what the glycemic index and glycemic load are, then you might want to skip this post unless you’re just here for the scintillating writing.

If I were to bring you into my office while you were fasting and check your blood sugar, then check it again every 15 – 30 minutes over the next two hours, I would find that your blood sugar levels wouldn’t change much.  Your blood sugar would remain at about, say, 85 mg/dL over the entire two hours.  Now, suppose I bring you in fasting, measure your blood sugar, then give you a piece of cake.  You eat the cake and I measure your blood sugar over the next two hours.  Your blood sugar would rapidly rise, then fall slowly, and return (assuming you’re not diabetic or glucose intolerant) to around your normal 85 mg/dL.

Scientists have known for years that normal blood sugars follow this kind of rapid increase, slow return to normal curve.  At some point someone asked the question: do different foods cause a different curve?  In other words, if someone eats a piece of cake does that make a different blood sugar curve than if that person eats a bowl of ice cream?

Let’s say we decide to find out how different carbohydrates affect you.  I draw your blood while you’re fasting, then give you 50 grams of glucose to drink.  I collect blood samples every 15-30 minutes for the next two hours and use that data to construct a curve of your blood sugar response to 50 grams of glucose.  A couple of days later I again draw your blood while you’re fasting, then I have you eat a piece of potato containing exactly 50 grams of potato starch.  Again I check your blood sugars over the next two hours and create a curve with your blood sugar data.

Now I’ve got a curve representing what happens to your blood sugar when you consume 50 grams of pure glucose and a curve representing what your blood sugar does when you eat 50 grams of potato carbohydrate.  I can compare these curves to see if there is any difference between the two.

When researchers look at any kind of data curve, they calculate the area under the curve, which, in the case of your blood sugar curve, represents the total amount of rise in blood sugar over time.  Assuming you have a normal blood sugar response to a carbohydrate load I would find that the area under the blood sugar curve after you had eaten the potato is about 80% of the area under the curve that we found after you had eaten the pure glucose.  Since the potato causes only an 80% increase in total blood sugar rise over the two hours as compared to the rise driven by the pure glucose we can say that the glycemic index of the potato is 80% that of glucose, or 80.

If we tried the same experiment with refried beans (like those you would find in Mexican food), we would find that the area under the blood glucose curve caused by the 50 grams of carbohydrate from the refried beans would be only about 40% of that caused by the pure glucose.  We would say that the refried beans have a glycemic index of 40.

The glycemic index figures you see published are derived the same way as I explained above.  Young, healthy subjects consume a test dose of glucose (or sometimes 50 grams of carb from white bread, which is also a common standard) and have their blood sugar curves measured.  They then come in on subsequent days and consume the test foods (the potato, beans, tomatoes, whatever) and have their blood sugar checked over two hours to create the blood sugar curves driven by the carbs in these foods.  These curves are compared to the pure glucose curves and a glycemic index is derived.

As you might imagine, it requires an enormous number of subjects and a whole lot of testing to come up with the long lists of glycemic indices for all the various foods.  It required herculean efforts by a whole lot of data gatherers to come up with these lists, lists I believe have very little value.  Before we get into that, however, let’s look at why the researchers who have spent so much time collecting this data think it’s important.

It would seem to make sense that one would be better off eating carbohydrates that are low-glycemic carbohydrates and raise blood sugar less than carbs that are high-glycemic carbs.  Wouldn’t it be better to eat refried beans with a glycemic index of 40 than to eat a potato with a glycemic index of 80?  Of course it would.  And if one is comparing carbs based solely on their glycemic indices, I would always recommend the choice with the lowest glycemic index.  But, are a bunch of low-glycemic index carbs better than very few carbs at all?  I don’t think so.

I think the whole glycemic index idea is interesting, but flawed as a tool for those seeking weight loss and/or better health.  Here’s why.

Glycemic index figures are calculated using measured doses of certain foods compared to pure glucose.  We know that 50 grams of potato starch gives 80% of the blood sugar rise compared to glucose.  But what happens if we eat some butter along with the potato?  Will the glucose curve be larger or smaller than it is with the potato alone?  No one knows.  Could you imagine the complexities involved in just trying to figure that one out?  How about if we eat the potato with butter and drink a cup of coffee?

Again, no one has a clue.

It’s nice to know what the glycemic index is for a potato all by itself, but since most of the meals we eat are a mixture of foods, how do we calculate the glycemic index for the whole meal?  It’s impossible.

And glycemic index figures are calculated using data derived from young, healthy subjects.  What if I’m old and have a touch of the Metabolic Syndrome?  Will these figures be the same for me?  I don’t think so.  But who knows?  It’s never been tested.

If I take in 50 grams of carbohydrate as potato or sugar or anything else, it’s probably not going to affect me that much.  What if I take in 300 grams?  I might (and probably would) react much differently.  So what good is an index derived from multiple studies of only 50 grams of carb, when I will more than likely eat much more than that?

Another problem with using the glycemic index is that fructose has a low glycemic index.  50 grams of pure fructose gives a blood sugar curve that is about 22-24% of that produced by glucose, which means that fructose has a glycemic index of only 22-24, which is very low by anyone’s estimation.  Why is fructose low?  Because it doesn’t convert to blood sugar, it converts to fat in the liver instead.  And causes a lot of problems.  But has a low, low glycemic index.  So foods containing fructose then have a low-glycemic index.

So, we don’t have a clue as to what the glycemic index of a mixed meal is.  We don’t know if our own personal glycemic index is the same as that of the young, healthy people used to create the lists we all use.  We don’t’ know if the glycemic index is the same with amounts of carb much larger than the 50 gram test loads.  And we may be encouraged to eat foods containing a lot of fructose because fructose has a low-glycemic index.  See why I have a problem with the concept?

Another issue that I touched on earlier is the difference between a lot of low-glycemic index carbohydrates and a small amount of high-glycemic carbohydrates.   If the glycemic index of refried beans is 40, can I eat twice as much of those as I can potato with a glycemic index of 80?  Or can I eat as much low-glycemic index stuff as I want without consequence because, after all, it won’t raise my blood sugar by much?

People who think the glycemic index is a good thing decided to take it a step further and came up with the concept of the glycemic load.  The glycemic load is the glycemic index multiplied times the grams of carb in the food in question.

Let’s look at a few foods.  Pure glucose has a glycemic index of 100 so if you eat 20 grams of glucose you have a glycemic load of 20. ( 1 (glycemic load converted to a fraction) X 20 grams = 20)  If we eat 20 grams of carbohydrate as refried beans we have a glycemic load of 8 (0.4 X 20), and if we eat 20 grams of carb as a potato we would have a glycemic load of 16 (0.8 X 20).  Theoretically, we could eat twice as much refried beans as we could potatoes and still have the same blood sugar changes.  But, unless we eat the refried beans while we’re fasting, and eat them all by themselves, and make sure we don’t get more than 50 grams of them, we really don’t know.  If we eat the refried beans along with an enchilada all bets are off.

By my thinking, it’s a whole lot easier to just count carbs and not worry about the glycemic index.  And it’s really a whole lot easier to simply count the carbs than it is to go to the trouble to look up the glycemic index, convert to a fraction, then multiply times the grams of carb.

All in all, the whole idea of the glycemic index and glycemic load is that by limiting the diet to lower glycemic index carbohydrates one can eat more carbohydrates and not suffer the problems that more carbohydrates typically cause.  Which, were it true, would be a good thing since most everyone (yours truly included) likes to eat carbohydrates.

But the paper I wrote about yesterday that was presented at the ASBP meeting showed that a plain old low-carb diet provided better results (for diabetics, at least) than a low-glycemic load diet.

Make it easy on yourself, and just count the grams of carb and be done with it.


  1. Does the high glycemic food require the same amount of insulin as a low glycemic food with the same amount of carbs? It seems like you still have to process those carbs, even if they don’t immediately raise your blood sugar (in other words, the low glycemic food wouldn’t raise it as much but for a longer period of time, requiring the same amount of insulin).
    Hi Ryan–
    You’re getting into the whole subject of the insulinogenic index, which is different.  Different foods may not raise blood sugar because they increase insulin enough to keep the blood sugar low.  As with the glycemic index, there are tables showing the insulinogenic index of various foods.  This index, in my opinion, suffers all the same problems as the glycemic index, i.e., what happens with mixed meals?, are all subjects the same?, etc.

  2. Thanks for reminding me of why I love low carb diets vs. low glycemic load diets. All the fruit and dried legumes allowed in low glycemic load diets did nothing but mess with my blood sugar and leave me feeling awful.
    I’m back to good ole’ carb counting and loving it.
    Thank you for educating us (once again) about what really works!
    Hi Sheryl–
    Thanks for reminding me.  I should have put in the post that another reason I don’t like the glycemic index is that many of my patients have tried to use it to increase the amount of carb in their diets without consequences and had the same experience as you.  If diabetic, their blood sugars went awry; if trying to lose weight, their weight loss stopped.  I haven’t found anything better than plain old carb counting.

  3. Thanks!!! I’ve tried to explain to many people why low carb is better than low GI!!!
    I’m very sensitive to carbs, both the total and the type. Many foods that are said to be low GI (mainly fruits) cause intense cravings in me!!! I know I’m insulin resistant, so I guess that’s part of the reason?!?
    Hi Cindy–
    You are not alone.  Many of our own patients have had the same problems with low GI foods, which is one of the reasons I began suspecting that the GI didn’t live up to its billing.
    Thanks for writing.

  4. Thanks for another great article. I would have also liked to hear more thoughts about how the insulin being produced to counteract the blood sugar increase is affected by these curves. I mean, I’m assuming that one could look at the area under the curve and get a numerical value that might indicate that while the blood sugar goes up only gradually, it also decreases gradually and so the net volume may not be that different from a high glycemic food that has a similar number of carbs? So in effect, your body is still producing the same amount of insulin, but just over a longer period of time. And if this is the case, is there any evidence that suggests that it doesn’t matter how fast the insulin is produced, but rather just how much of it is getting produced over time?
    Hi Levi–
    Take a look at the answer to Ryan’s comment above.  He asked the same question.

  5. I’ve never paid much attention to the index, though I do have one diabetic acquaintance who says it’s working for her. Wouldn’t work for me. Carbs and me do not mix and I’m not even diabetic!
    Hi Victoria–
    Thanks for the feedback.

  6. Hello there,
    Wondering what your take is on the study you posted on your site. Specifically the finding that there was a 30 percent lower risk with veggie fat intake vs, animal fat intake.
    Hi Daniel–
    I’m working on a long post on that very subject right now.  As soon as I figure out how to insert the charts I need I’ll have it posted.

  7. After two gestational diabetic pregnancies, I am obviously at high risk for Type II. My PCP directed me to a low GI diet. It translated to a higher A1C for me. When I questioned why I needed carbs he said that a low GI diet will help sensitize my cells to insulin and without carbs, there’s nothing to sensitize to. Is that true?? Except for oats and some fruit, I’m back to good old low carb and feeling so much better.
    Thanks for your blog. I enjoy reading it!
    Hi Cheryl–
    I’m glad you enjoy the blog.  I’ve had the same experience that you describe with many of my diabetic patients.  I gave the low GI diet a try when it first appeared in the medical literature, but for my patients, it didn’t live up to its billing.

  8. Alcohol consumption and brain shrinkage !
    Hi Simon–
    I read the article.  Sounds like dueling PhDs to me.  There are a ton of other article supporting the opposite view, so I don’t get all wrapped up in one contrary report.  I find it hard to believe that “low to moderate” drinking can cause such problems.  But, I’m open minded.  If the evidence continues to amass, I may amend my opinion.  Even so, I’ll probably opt to live with a smaller brain rather than abandon my low to moderate consumption of alcohol.

  9. Hi sorry if you’ve already received this question but I’m having trouble with sending comments (it doesn’t seem to like me!)
    In Australia the measures for blood sugar are different. Can you provide some comparisons? My husband has been testing after some health issues. He finds that first thing in the morning he is high (7 to 8.5) but later in the day after he has eaten it is normal (5 to 7) -is this normal? He has been trying to eat low carb for a few months & is going quite well – has lost about 12kg with blood pressure lower but still not normal. His Doc wanted him to do a glucose tolerance test but we freaked when we saw what you had to eat for 3-4 days and said no way! (Especially as after this they would only say “watch your diet” anyway!)
    Hi Lynne–
    I could have sworn that I answered this comment, and it turns out that I did. You posted it under the Spoonful of Sugar post that I wrote a while back.
    I’ve copied my response and pasted here as well.
    Hi Lynne–
    You’re way of measuring blood sugar is in mmol/L. You can convert to our old fashioned mg/dL by multiplying your figures by 18 or conversely can convert ours to yours by dividing by 18.
    Your husband is experiencing what is called the dawn phenomenon. The liver produces and releases glucose into the blood unless it is turned off by insulin. The liver tends to make more overnight to compensate for the fact that one doesn’t usually eat while sleeping, yet one’s blood sugar level needs to be maintained. The liver produces enough glucose to keep the blood sugar at a normal level. If the level gets a little high, the pancreas releases a little spurt of insulin that signals to the liver to quit producing sugar for a while. An insulin resistant person’s liver often doesn’t get the signal and keeps on producing sugar leading to elevated sugars in the morning that tend to go down later in the day.
    In my experience, the best treatment is a low-carb diet over the long haul to re-establish insulin sensitivity.
    Hope this helps.

  10. Dr Eades,
    Thank you for the article. As you point out, Glycemic Load too has its limitations. However, I would highly recommend reading “The Effects of Fat and Protein on Glycemic Responses in Nondiabetic Humans Vary with Waist Circumference, Fasting Plasma Insulin, and Dietary Fiber Intake” (2006) by Dr. Wolever et. al. The paper shows how GL can be adjusted to take into account all proteins, carbohydrates and fats consumed to arrive at an adjusted GL for that meal.

    1. I’ve read that paper. As I recall, there we 20 subjects tested 20 times or something like that, which gets it a little more refined, but it still doesn’t say what happens when you’re 60 vs 20, or when you’re fat vs thin, or what happens due to a hundred other variables.

Leave a Reply

Your email address will not be published. Required fields are marked *