The answer is the reason this chapter exists. The glycemic index is a useful idea, but it answers only one question: how fast does the carbohydrate in this food reach your blood? It does not say how much carbohydrate you are actually eating, and it does not say what the rest of your meal, your cooking, or your own body will do to the answer. Once you see what the index can and cannot tell you, it stops being a list of forbidden foods and becomes what it was meant to be: one tool among several for choosing carbohydrates well.

We take it in three steps. First, what the glycemic index measures and how. Second, the glycemic load, which adds the missing piece, the portion. Third, what the research says about using both in real life, why the numbers you read in tables will not always match what your own meter shows, and what you can do about it.

5.1 What the Glycemic Index Measures

A ranking by speed

The glycemic index (GI) ranks carbohydrate-containing foods by how much they raise blood glucose in the two hours after eating, compared with the same amount of pure glucose. Pure glucose is given a value of 100 by definition. A food with a GI of 50 raises blood glucose about half as much as the same amount of glucose would; a food with a GI of 90 raises it almost as much.

Because the index is about carbohydrate, foods that contain almost none, such as meat, fish, eggs, oils, and most non-starchy vegetables, do not have a meaningful GI at all. The index is also not a measure of how healthy a food is. It says something about glucose, and nothing about vitamins, fibre, salt, or what a food does to your heart.

How a GI value is measured

In a laboratory, a group of volunteers eats a test food containing a fixed amount of available carbohydrate, usually 50 grams (in simple terms, the carbohydrate your body can digest and absorb, which is roughly total carbohydrate minus fibre). Their blood glucose is measured over the next two hours. On another day, the same people eat pure glucose containing the same 50 grams of carbohydrate, and their glucose is measured in the same way. Each person's response to the test food is compared with their own response to glucose, and the results are averaged across the group. An international standard, ISO 26642, describes this method and recommends how to classify foods.[@iso26642]

{{fig:F5-A}}

Science Corner: How a GI number is calculated For each volunteer, a laboratory draws the blood glucose curve over two hours and measures the area under the curve that lies above the starting (fasting) level. The GI for that person is the area after the test food divided by the area after glucose, multiplied by 100. The published value is the average of those individual results across the volunteers.[@iso26642] Two consequences follow. First, the index is a relative number, a ratio and not a rise in mg/dL. Second, it describes an average across a small group of healthy volunteers, not the response of any one person, a point that matters greatly in section 5.3.

The three bands

By the classification recommended in the ISO standard, foods are grouped into three bands: low GI (55 or less), medium GI (56 to 69), and high GI (70 or more).[@iso26642] Foods such as most legumes, many intact whole grains, and most fruits and vegetables tend to fall in the low band. White bread, some white rice and potato products, and many refined breakfast cereals tend to fall in the high band. These are patterns, not rules: as you will see, the same food can sit in more than one band depending on variety, ripeness, and cooking.

Where to find GI values

The most complete resource is the International Tables of Glycemic Index and Glycemic Load Values, last updated in 2021, which lists more than 4,000 items collected from published studies. The 2021 edition added, among others, Asian fruits such as lychee, dragon fruit, and pomelo, and separates values obtained by the ISO method from those obtained by other methods.[@atkinson2021] If your staple food is not in the tables, that is normal: many traditional dishes have never been tested, and a value borrowed from a similar-looking food may be a poor guide.

5.2 From Index to Load: Adding the Portion

The missing piece

Return to the watermelon and the doughnut. The index tells you the speed of the carbohydrate; it says nothing about the amount in the bite you actually eat. Watermelon is mostly water. A typical serving contains only about 11 grams of carbohydrate, while a medium doughnut contains about 23 grams.[@lpi2016] Same speed, but about twice the quantity.

That is what the glycemic load (GL) captures. It multiplies quality (the GI) by quantity (the grams of available carbohydrate in the serving):

GL = GI × grams of available carbohydrate in the serving ÷ 100

One unit of GL is roughly the effect of one gram of pure glucose. For single foods, a GL of 10 or less is considered low, 11 to 19 medium, and 20 or more high.[@lpi2016]

{{fig:F5-B}}

Worked examples

For the watermelon: 76 × 11 ÷ 100 = about 8, a low load. For the doughnut: 76 × 23 ÷ 100 = about 17, a medium load and more than double.[@lpi2016] The next table uses simple round numbers to show how the same GI behaves at different portion sizes. The GL values were calculated from the formula above.

The foods in this table are imaginary and the numbers are chosen only to show the arithmetic; they are not measurements of any real food.

Now compare two different foods at the same portion. A food with a GI of 40 and 30 grams of available carbohydrate has a GL of 12, medium, while the GI 70 food with the same 30 grams has a GL of 21, high. So a low-GI food can still deliver a large load if the portion is big enough, and a high-GI food can be modest in load if the portion is small. Both quality and quantity matter, and neither alone tells the whole story.

Why two tables can give you two answers

If you look up watermelon in a second source, you may find a different number. One widely read health publication, for example, lists watermelon with a GI of 80 and a glycemic load of only 5.[@harvard2023] The direction of the message is the same: high index, low load. But the specific numbers differ from those above, because different tables use different studies, different serving sizes, and different measurement methods. The 2021 international tables tried to reduce this problem by standardising carbohydrate portions (between 15 and 30 grams depending on the food group) for the load calculation.[@atkinson2021] The practical lesson: treat any GI or GL figure as an approximate guide, not as a precise property of a food.

Science Corner: Is total carbohydrate a better guide than GI or GL? This is a real debate. The Harvard Health resource cited above notes that the total amount of carbohydrate in a food, rather than its GI or GL, is a stronger predictor of the blood sugar response, and some dietitians find GL-based eating unnecessarily complex.[@harvard2023] The glycemic load already includes the amount of carbohydrate, so the two approaches overlap. A sensible middle path, which the next chapters follow: use the amount of carbohydrate as your first guide (Chapters 6 and 7 show how to build plates around it), and use the type of carbohydrate, and the GI as one way of judging that type, to choose between foods that are otherwise similar.

5.3 Using GI and GL in Real Life

What the research says

Does choosing lower-GI foods actually help people with diabetes? A large systematic review and meta-analysis of randomised trials in people with type 1 and type 2 diabetes pooled 27 reports (29 trial comparisons) and 1,617 participants, with a median follow-up of 12 weeks. On average, low-GI or low-GL dietary patterns lowered HbA1c by 0.31 percentage points (95% confidence interval 0.19 to 0.42) compared with control diets. They also modestly lowered LDL cholesterol, triglycerides, body mass index, and C-reactive protein (a marker of inflammation), without clear changes in HDL cholesterol, waist circumference, or blood pressure. The authors concluded that low-GI/GL patterns can be recommended in diabetes management, while noting that larger, higher-quality trials with bigger differences between the diets are needed.[@chiavaroli2021]

Science Corner: How big is an HbA1c drop of 0.31? Recall from Chapter 4 that the average glucose estimated from HbA1c changes by about 28.7 mg/dL for each 1 percentage point.[@nathan2008] A 0.31-point drop therefore corresponds to an average glucose roughly 9 mg/dL (about 0.5 mmol/L) lower. That is a modest but real change, and it came from food alone. It is also an average across studies that lasted, at the median, only 12 weeks, so it tells us little about what happens over many years. The right way to read it: helpful, not miraculous.

The European Association for the Study of Diabetes reached a similar position in its 2023 evidence-based dietary recommendations. It judged that low-GI or low-GL diets can improve glycemia and some cardiometabolic risk factors in people with diabetes, with moderate certainty of evidence, but presented this as one useful marker of carbohydrate quality among several, not as the main rule. Its emphasis was on minimally processed plant foods, at least 35 grams of fibre a day from whole grains, vegetables, legumes, nuts, seeds, and whole fruit, and on intact whole grains over finely milled ones, because processing and milling appear to reduce whole grains' ability to lower glycemia.[@easd2023]

Why the table will not always match your meter

If you have ever eaten a "low-GI" food and watched your glucose climb, you were not imagining it. Three things explain the gap.

The same food is not always the same food. GI values are affected by the variety of the food, its ripeness, how it is processed, and how it is cooked.[@lpi2016] Rice is the clearest example. A systematic review of 32 randomised trials found that the GI of rice varieties ranged from 48 to 93. The main factor was the ratio of two starches, amylose and amylopectin: rice higher in amylose produced lower responses. Pressure parboiling lowered GI by about 30 percent in the studies reviewed. And, surprisingly, brown and white rice did not differ consistently when prepared with realistic cooking times, and milling and polishing had inconsistent effects on the blood glucose response.[@boers2015] For someone whose main food is rice, the type of rice and how it is prepared may matter more than the colour.

A meal is not a single food. The table values come from testing one food on its own. In a real meal, fat, protein, and fibre change how fast the carbohydrate is absorbed. Fibre and fat generally lower a food's GI, and the more processed a food is, the higher its GI tends to be.[@harvard2023] A bowl of rice eaten with beans, vegetables, and a little oil is a different experience from the same rice eaten alone.

You are not the average volunteer. This is the least appreciated point. In one study of commercial white bread, individual GI values ranged from 44 to 132 across the volunteers, and even the same person's values varied from one test day to another.[@vegalopez2007] A larger study of 63 healthy adults found that GI values could vary by about 20 percent within the same person and by about 25 percent between people, and that white bread could land in the low, medium, and high bands depending on the person and the day.[@matthan2016] A separate study that followed 800 adults with continuous glucose monitors, and measured the response to nearly 47,000 meals, found a very high variation between people in their response to the same foods, including bread. A computer model that combined personal factors (such as age, body mass index, blood tests, activity, sleep, and gut bacteria) with what was eaten predicted the responses better than carbohydrate content alone.[@zeevi2015]

None of this means the GI is useless. The authors of the white bread study noted that GI may be reasonable as an indicator for groups, while there is still considerable uncertainty when it is applied to individuals.[@vegalopez2007] It means the table is where you start, and your own measurements are where you finish. This is exactly why Chapter 4 introduced the numbers, and why Chapter 12 shows how to use a meter or a continuous glucose monitor to check your own responses.

Practical ways to use the ideas

Bringing this together, here are steps that fit almost any cuisine.

Start with whole, minimally processed foods. Legumes such as lentils, chickpeas, and beans, intact grains such as barley, oats, and whole-grain rice, non-starchy vegetables, and whole fruit tend to release carbohydrate more slowly than refined flours and sugary drinks, and they bring fibre with them.[@easd2023]

Watch the portion of starchy foods, not only their type. Because load is index times amount, halving a large portion of a medium-GI food can lower the load more than switching foods.

Pay attention to how the staple is prepared. In one small crossover trial in 15 healthy adults, white rice that was cooked, cooled for 24 hours at 4 °C, and then reheated produced a lower glucose response than freshly cooked rice (125 versus 152 mmol·min/L), and the amount of resistant starch, a starch that resists digestion, rose from 0.64 to 1.65 grams per 100 grams.[@sonia2015] This is a small study in people without diabetes, so treat it as an interesting idea to test with your own meter, not as a guarantee. A note of caution comes from a randomised crossover study in 32 people with type 1 diabetes on insulin pumps. Cooled and reheated rice did lower their post-meal glucose, but there were more hypoglycaemic episodes (12 versus 3) because their usual insulin doses were unchanged.[@strozyk2022] If you use insulin or medicines that can cause lows, please talk to your care team before changing how you prepare your staples.

Do not let a low GI excuse a poor food. Because fat lowers a food's GI, a high-fat, high-sugar food can look "low GI" in a table.[@harvard2023] GI is one lens on carbohydrate; it does not replace looking at the overall quality of what you eat.

Test, do not guess. If you use a glucose meter, you can check a new meal by measuring before eating and one to two hours after the start of the meal. The ADA suggests a peak after-meal glucose of below 180 mg/dL (10.0 mmol/L) for many non-pregnant adults, a target to be individualised with your care team, as we saw in Chapter 4.[@adaguide2023] If a "safe" food repeatedly pushes you above your target, it is not the right food for you at that portion, whatever the table says. If a "forbidden" food, like watermelon in a sensible slice, keeps you comfortably in range, the table was too blunt for your body.

Science Corner: Should I follow a "low GI diet"? A low-GI diet is not a separate medical treatment; it is a way of picking carbohydrate foods. The evidence supports it as a modest, useful option, especially when it steers you toward legumes, whole grains, vegetables, and fruit, but it is not required, and it is not the only route to good glucose control. Other patterns that reduce refined carbohydrate and increase fibre, discussed in Chapter 6, share many of the same benefits. What matters most is that you can keep eating this way for years.

Key Takeaways

• The glycemic index (GI) ranks carbohydrate foods by how much they raise blood glucose compared with pure glucose (100). By the ISO classification, low is 55 or less, medium 56 to 69, high 70 or more.
• GI shows speed, not amount. The glycemic load (GL = GI × available carbohydrate in grams ÷ 100) adds the portion. Watermelon has a high GI but a low GL; a doughnut has the same GI and a larger load.
• GI and GL values in tables are approximate and differ between sources. Variety, ripeness, processing, and cooking all change them. The GI of rice varieties, for example, ranges from 48 to 93.
• Low-GI and low-GL diets lowered HbA1c by about 0.31 percentage points on average in a meta-analysis of randomised trials, a modest but real benefit judged of moderate certainty by European experts.
• Responses to the same food vary widely between people and even in the same person. The table is a starting point; your own meter or sensor gives the final answer.
• Choose whole, minimally processed carbohydrate foods, watch your portions, and do not treat "low GI" as a synonym for "healthy".

Action Points

1. Pick one starchy staple you eat every day (rice, bread, potatoes, maize, cassava, or noodles). Look up its GI in a reliable table, and note whether the variety or the way you cook it could change it.
2. Practise the GL formula once. Choose a food you eat, find its GI and grams of available carbohydrate in a serving, and calculate the GL. Compare it with a food you consider "forbidden".
3. Add fibre to your main meal. Put legumes, vegetables, or a whole-grain option beside your usual starch, and see how you feel and (if you measure) how your glucose responds.
4. Test one meal twice. If you have a meter, measure before and one to two hours after the start of the same meal on two different days, and write both numbers down.
5. If you use insulin or medicines that can cause lows, ask your care team before making big changes to the type of carbohydrate you eat or how you prepare it.

This book is intended for education and does not replace personal medical advice. If you have diabetes or take glucose-lowering medication, please consult your healthcare team before changing your diet, exercise, or treatment.