Low Insulin Load Foods: The Complete List (2026)

The lowest insulin load foods are oils, non-starchy vegetables, avocado, nuts and seeds, eggs, and plain meat, fish and tofu. Each has almost no net carbohydrate, so its Insulin Load Score is mostly under 5 per serving (under 7 for the lower-load fruit). One caveat applies to the protein foods: lab measurements show some of them raise insulin more than their carb content predicts. The full list below gives a score for every food, grouped by type.

By Naor Lugassi, founder of Kcaly AI. Last updated: November 4, 2026. Not medical advice, and not a tool for insulin dosing.

How to read the scores

The scores below use the Insulin Load Score (ILS) formula from our insulin load calculator: 0.56 × net carbs + 0.1 × protein + 0.03 × fat, in grams per serving, with fibre subtracted from carbs. Our calculator calls 0 to 30 “low”; every food on this list is far below that, so we show the decimal to let you compare. For the theory, see insulin load score and insulin load score explained.

Two honest limits. First, the serving sizes differ (100 g of broccoli versus 28 g of almonds), so a score is per the serving shown, not per calorie. Second, the ILS is a simplified estimate. It is inspired by the Food Insulin Index, which was measured in people, but it has not been validated against those measurements.

Dairy and a few staples worth knowing

Hard cheeses, plain Greek yogurt and cottage cheese score low on the formula (roughly 3 to 5 per serving shown in our Greek yogurt and cottage cheese pages). Milk is different: a cup of whole milk lands near 8 because of its lactose, and dairy is one of the food groups where measured insulin tends to run higher than the carb content suggests.

Why a low score is not the whole story

The original measurement work is the Food Insulin Index from the University of Sydney (Holt, Miller and Petocz, American Journal of Clinical Nutrition, 1997; PubMed). They fed 38 foods to healthy volunteers in equal-energy (1000 kJ) portions and compared the insulin response with white bread. Their finding that matters here: protein-rich foods and bakery products produced insulin responses that were disproportionately high compared with their glucose responses.

In practice that means eggs, meat and fish are not “zero insulin” foods, and the macro-based score understates this. If you manage diabetes, your own glucose readings and care team outrank any list, including this one.

Foods that are moderate, for contrast

These sit above the lists but are not “high” either. They are the foods to portion rather than avoid.

Refined starches are higher still: a plain bagel scores about 31 and granola about 34 per serving in our data. Browse any food on the insulin load of foods pages.

Using the list in real meals

Meals, not single foods, are what you eat. A plate of salmon, broccoli and olive oil scores a few points in total; swap the broccoli for white rice and the carbs take over. The practical move is to build around the low-load groups and portion the starch, rather than eliminate it. If your goal is weight, calories still decide the outcome, so pair this with a TDEE estimate.

One thing we see in how people use Kcaly AI: the people who care about insulin load rarely want to look foods up one by one. They log the meal by photo, text or voice note on WhatsApp and read the score off the reply (an observation from our own use, not a study). If that is you, the Insulin Load Score feature shows it per meal, and our calorie tracker for diabetes page explains the setup. Or start the 3-day free trial.

Sources

• Holt SHA, Miller JC, Petocz P. An insulin index of foods: the insulin demand generated by 1000-kJ portions of common foods. Am J Clin Nutr 1997;66:1264-76.
• Nutrient values per serving: USDA FoodData Central (fdc.nal.usda.gov), via the Kcaly AI food database.

Frequently asked questions