Food guide · Type 1

A carb isn’t a carb

Knock knock.Who’s there?40 grams of carbs.40 grams of carbs who?

Same 40 grams at the door, a different guest every time. Open each card to see who it is and how it hits.

Coll. Haley Maddox · Oct 2026

Just a reminder
My diabetes care is managed in close partnership with my care team. This page shares what works for me. It isn't medical advice. Talk with your own doctor or diabetes care team before you change anything about your care.

Start here

GI vs GL

This builds on the bolusing guide, which covers how much to count and when. Here's the next layer: a carb isn't just a carb.

Both numbers describe carbohydrate foods. The glycemic index is about the kind of carbohydrate. The glycemic load adds the amount you actually eat.

Glycemic index (GI) How much a food raises blood glucose over two hours, compared with pure glucose, when both give the same amount of carbohydrate. Glucose is 100.

Low: 55 or less
Medium: 56 to 69
High: 70 or more

Glycemic load (GL) The GI scaled to a real serving. It tells you how big the effect of that portion is likely to be.

Low: 10 or less
Medium: 11 to 19
High: 20 or more

The formula

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

Watermelon: a high GI of 72, but a 120 g serving holds only a little carbohydrate, so its GL is 4. Low.

GI bands are from the 2021 international GI tables.¹ GL bands are from University of Illinois Extension.² Watermelon is from the Harvard Health table of University of Sydney values.³

Measure

The test

A GI comes from people, not a lab machine. Volunteers fast overnight, eat a portion of the food that holds 50 g of available carbohydrate, and have their blood glucose checked over the next two hours. On another day they drink 50 g of glucose. The GI is the area under the food's curve as a percentage of the area under the glucose curve.

GI = test food area ÷ glucose area × 100.

David Jenkins and his team at the University of Toronto introduced the idea in 1981. They fed 62 common foods to groups of 5 to 10 healthy fasting volunteers. Foods with more fat and protein tended to have a lower GI; fiber and sugar content showed no clear link.

Most GIs are measured in people without diabetes. When the same 20 staple foods were tested in both groups, the results lined up closely (r = 0.94), so the rankings still apply to people with diabetes.

From Jenkins et al., 1981,⁴ and Atkinson et al., 2008.⁵

Compare

Examples

Here are common foods placed by both numbers. Right means a faster, higher rise per gram of carbohydrate. Up means a bigger effect from a normal serving. Watermelon sits far right but low; long-cooked spaghetti sits in the middle but high.

Tap or hover a dot for its values. Higher-fat foods are marked in orange. Swipe sideways to see the whole chart.

Glycemic index and load of common foods
Food	GI	GI band	Normal serving

See it in action with insulin. The foods marked In the tool are the five in Learn now, bolus later, where you can slide Humalog or Afrezza against each one.

Why two ice creams? Premium ice cream has more fat than regular, and fat slows the sugar. Same food, so the gap (61 vs. 37) shows what fat does on its own.
Why does juice have a low GI? Orange juice tests at 50 because part of its sugar is fructose, which barely raises glucose (GI 15 in the 2008 tables). It's still a drink, so it starts working quickly, which is why juice is a go-to for treating a low.

Good to know: a GI is a lab average for one food eaten alone. The same food can test differently by brand, recipe, and cooking time. White spaghetti boiled 5 minutes has a GI of 38; boiled 20 minutes, 61.

Values are from the Harvard Health table,³ drawn from the 2002 international GI tables (Foster-Powell, Holt, and Brand-Miller).⁶ "GI band" is low (55 or less), medium, or high (70 or more). "Normal serving" is the glycemic load band for the portion each table lists.

Highs

Fat, too little insulin, or both

A rise after a meal usually has one of two causes, and sometimes both. Rapid-acting insulin like Humalog starts working in 15 to 30 minutes, peaks at 1 to 2 hours, and lasts 4 to 6 hours. What matters is whether that matches when the carbs arrive, and whether there was enough of it.

Late rise from fat. The dose was right, but the meal's fat slows its carbs (more on that below). The insulin peaks first, glucose can dip, and then the pizza keeps arriving after the insulin is mostly done.
Not enough insulin. A missed or short bolus. Glucose climbs within the first hour or two and stays up.
Both. A short bolus for a high-fat meal. It climbs early and keeps climbing late. In one study, a high-fat dinner needed more insulin than a low-fat one with the same carbs and still ran higher.

Starts fine or even dips, then climbs 3 or more hours after eating, once the insulin is mostly done.

Swipe sideways to see the whole chart.

How to tell the causes apart
Cause	First 2 hours	When it goes high
Late rise from fat	Flat or a dip	3 or more hours after eating
Not enough insulin	Climbs right away	Within 1 to 2 hours, and stays
Both	Climbs right away	Early, then higher still later

Researchers have tested splitting the same total dose for pizza. In one study of 24 people with type 1 on pumps, giving half up front and half over 8 hours gave the lowest glucose 8 to 12 hours after dinner, without more lows. How to split a dose is a decision to make with your care team.

Try these highs yourself. In Learn now, bolus later, pizza with Humalog or Afrezza shows the late rise. Then split the dose, or slide the timing earlier or later, and watch what changes.

Insulin timing is from the ADCES insulin types chart.⁷ The high-fat dinner needing more insulin is from Wolpert et al., 2013.⁸ The early dip and late rise are from Bell et al., 2015.⁹ The pizza study is Jones et al., 2005.¹⁰

Fat

How fat slows carbs

Fat slows how fast the stomach empties into the gut, so the sugar in a meal reaches the blood more gradually. That's why higher-fat versions of the same food test at a lower GI.

80 → 36Pizza: plain dough with tomato and parmesan, then a supreme loaded with meat and cheese

61 → 37Ice cream: regular, then richer premium

Chocolate, fries, and other fatty foods follow the same idea. In the 2008 tables, ice cream across all studies averaged 51, and chocolate 40.

Slower is not the same as smaller

A low GI only describes the first two hours. Research in type 1 diabetes shows the fat can push glucose up later. A 2015 review from the University of Sydney found that high-fat meals mainly caused late high glucose, sometimes after lower readings in the first two to three hours.

In one trial, 33 children and teens with type 1 ate pancake breakfasts with the same 30 g of carbohydrate but either 4 g or 35 g of fat. From about 3½ to 5 hours, glucose after the high-fat breakfast sat about 32 mg/dL above where it started, compared with about 9 mg/dL below for the low-fat one. Adding protein on top raised it further.

Pizza and ice cream are from the Harvard Health table.³ Averages for ice cream and chocolate are from Atkinson et al., 2008.⁵ The review is Bell et al., 2015.⁹ The pancake trial is Smart et al., 2013 (mmol/L × 18 = mg/dL).⁸

Lows

Lows to highs

A low is under 70 mg/dL. The fastest carbs fix it best, and fat works against it. That same timing explains why a low can turn into a high. For how to treat a low, see Type 1 questions.

Why a low can turn into a high

Slow carbsChocolate, cookies, and other baked goods have fat that slows how fast the sugar gets in, the same effect described above. At the next check you can still be low, so you eat more. Then it all arrives at once.

The sensor runs behindA CGM reads the fluid between cells, not blood. In one study the raw signal ran about 9 to 10 minutes behind blood glucose. The sugar may already be working while the number still says low. A fingerstick shows where you are now.

Eating to the feelingLows bring strong hunger and shakiness, and waiting for the sugar to work feels long. Eating until the feeling passes can mean far more than the low needs. Researchers note that overtreating can lead to bigger swings.

Your body helps tooA low sets off hormones such as epinephrine that tell the liver to release stored glucose. Most people with type 1 lose the glucagon part of this response, but the others can still add to what you ate.

Swipe sideways to see the whole chart.

Fat and absorption are from the CDC.¹¹ Sensor delay is from Schmelzeisen-Redeker et al., 2015.¹² Overtreating is from Martyn-Nemeth et al., 2019.¹³ Hormones are from Diabetes Self-Management.¹⁴

Review

Research

1981

The first GI

Jenkins and colleagues ranked 62 foods by the two-hour glucose rise in healthy volunteers. Legumes and dairy were lowest; vegetables and breakfast cereals highest.

Jenkins et al., Am J Clin Nutr 34:362–366.⁴

2002–2021

The international tables

The University of Sydney group keeps the reference list. The 2021 edition has over 4,000 items, about 2,100 tested by the ISO standard method, with GL worked out on standard carbohydrate portions.

Atkinson et al., Am J Clin Nutr 2021.¹ 2008 edition.⁵

2005

A pizza night study

24 people with type 1 ate the same pizza dinner on different nights with the same total insulin given three ways. Spreading half the dose over 8 hours gave the lowest glucose 8 to 12 hours later.

Jones et al., Diabetes Technol Ther 7(2):233–240.¹⁰

2013

Fat in type 1

In 33 young people with type 1, adding fat or protein to the same carbs raised glucose 3 to 5 hours later, and the two effects added together.

Smart et al., Diabetes Care 36(12):3897–3902.⁸

2015

Fat, protein, and GI

A systematic review of meal studies in type 1. Fat, protein, and GI each changed the glucose curve after meals; fat's main effect was a late rise.

Bell et al., Diabetes Care 38(6):1008–1015.⁹

2015

Everyone is different

The Weizmann Institute tracked 800 people on CGM for a week across 46,898 meals. People's responses to identical meals varied widely, and a model using gut bacteria and personal data predicted them better than carbs alone.

Zeevi et al., Cell 163:1079–1094.¹⁵

What the numbers can't tell you

A GI is tested on one food eaten alone. A real meal mixes foods, and fat and protein change the curve.
It only covers two hours, so it misses the late rise from fat.
Brand, recipe, cooking time, and the person eating all shift the result. Your own CGM is the best test of your own response.
A low GI isn't a health label. The 2008 tables note that chocolate is low GI, but its saturated fat lowers its nutritional value.⁵

Sources