Illustration of a bowl of pan-browned tofu cubes with broccoli and cauliflower on a wooden board, beside a small dish of hemp seeds, a few almonds and half an avocado.

Can you be vegan and low-carb? Yes, and the trial behind it never got near keto

Gregory Knox
November 3, 2026
9 mins

Yes. In a 6-month trial, overweight adults with high cholesterol lost 1.1 kg more on a vegan low-carb diet than on a high-carb vegetarian one, and their LDL cholesterol fell further. In pooled cohort data, swapping carbohydrate for plant fat and protein was linked to lower mortality, and swapping it for animal fat and protein to higher.

That trial, nicknamed Eco-Atkins, is the only randomized vegan low-carb trial we read for this page, and its low-carb group never ate anything like keto. The diet was prescribed at 26% of calories from carbohydrate. After a first month in which all their food was provided, participants spent six months choosing their own, and over that phase the low-carb group averaged 39.6% of calories from what the paper calls available carbohydrate, against 53.9% in the comparison group. The paper doesn’t comment on the distance from 26%, and its intake figures fill in the people who dropped out by multiple imputation. For scale, a 2024 expert consensus puts ketogenic diets below 10% of calories and calls anything from 26% to 44% moderate-carbohydrate. Anyone who went low-carb by swearing off wheat should also know that more than half the protein prescribed in this trial came from wheat gluten, in nut bread and seitan products. If protein is your worry, our look at whether vegans run short of protein has the survey figures.

How low is low-carb, and where do vegans start?

The 2024 consensus, written by attendees of the Scientific Forum on Nutrition, Wellness, and Lower-Carbohydrate Diets, sets its bands on a 2,000-calorie day, which turns the percentages into grams. Ketogenic means 20 to 50 g a day, low-carbohydrate 50 to 129 g, and moderate starts at 130 g. The Forum was paid for by an unrestricted educational grant from Simply Good Foods, and several authors disclose ties to low-carb businesses or books, among them stock options or equity in Virta Health and royalties on low-carbohydrate titles.

The official numbers sit right on the line. The National Academies set the adult Recommended Dietary Allowance for carbohydrate at 130 g a day, exactly where the consensus’s moderate band begins, and their acceptable range for adults runs from 45% to 65% of calories. The consensus authors argue that the 130 g figure rests on the brain’s average glucose use and isn’t supported by current evidence, because the brain can also use ketones and the body can make glucose from other sources. Eat exactly the RDA and, by the low-carb camp’s own table, you are not on a low-carb diet.

Vegans start further up. In the Adventist Health Study 2, vegans got 58.1% of their calories from carbohydrate against 51.4% for nonvegetarians, with intakes standardized to 2,000 calories a day. That comes to about 290 g of carbohydrate a day (our arithmetic, at 4 calories a gram), so a vegan eating like that cohort would have to cut more than half of it to get under 130 g.

What the Eco-Atkins trial found

The trial ran from April 2005 to November 2006 at a hospital nutrition research center in Toronto, with overweight men and postmenopausal women whose LDL cholesterol was high. Of those, 39 went on to the six-month phase, split between the vegan low-carb diet and a high-carbohydrate lacto-ovo vegetarian diet, and both groups were encouraged to eat only 60% of their estimated calorie needs. On paper the low-carb diet was 26% carbohydrate, 31% protein and 43% fat, against 58%, 16% and 25% for the comparison diet.

The low-carb diet’s carbohydrate came mainly from viscous-fiber foods such as oats and barley, and from low-starch vegetables, okra and eggplant above all. Soy turned up as burgers, deli slices, breakfast links, veggie bacon, tofu and soy milk, and the nuts were almonds, cashews, hazelnuts, macadamias, pecans and pistachios. Veggie bacon, on a cholesterol trial’s prescription. Participants could buy the nut bread, which the paper calls “no” starch, at the research center.

Weight fell 6.9 kg on the low-carb diet and 5.8 kg on the high-carb one, a difference of 1.1 kg (95% confidence interval 0.0 to 2.1 kg, p=0.047). LDL cholesterol fell 0.49 mmol/L more on the low-carb diet and triglycerides 0.34 mmol/L more, while blood pressure didn’t differ between the groups.

The authors name the small sample and the high dropout rate as the study’s weaknesses. Half the low-carb group and 32% of the high-carb group didn’t finish, nobody could be blinded to their diet, and intake of the three cholesterol-lowering components averaged 33.6% of what was prescribed. The study was funded in part by Solae and by Loblaw Companies, and three of its authors were Solae employees. It tested one carefully built menu in people with high cholesterol, and its authors say how widely the diet applies remains to be determined.

Does it matter where the fat and protein come from?

In the cohort data, a great deal. Seidelmann and colleagues followed 15,428 US adults aged 45 to 64 in the ARIC study for a median of 25 years, during which 6,283 of them died. Mortality was lowest at 50% to 55% of calories from carbohydrate, and higher both under 40% and over 70%. Pooling eight cohorts with 432,179 people, they found mortality rose when carbohydrate was exchanged for animal fat or protein (hazard ratio 1.18) and fell when the exchange was plant-based (0.82). Compared with higher-carbohydrate diets, the animal-leaning low-carb pattern had more servings of beef, pork and lamb, and the plant-leaning one more servings of nuts, peanut butter and dark or grain breads.

This is observational data, and diet was measured by questionnaire at only two points across six years. The authors also say that relatively few people in the cohorts ate low-carb diets built mainly on plant protein and fat. Their conclusion is that animal-based low-carbohydrate diets should be discouraged, and that people who do restrict carbohydrate could consider replacing it with mainly plant-based fats and proteins.

The 2025 to 2030 Dietary Guidelines for Americans now say people with certain chronic diseases may do better on a lower carbohydrate diet, and tell them to work with a health care professional to choose one. The same guidelines list beans and lentils among the protein foods for vegetarians and vegans, which is where a vegan low-carb plan starts to pinch.

Meet 6 low-carb vegan foods

These are USDA’s figures per 100 g, rounded to one decimal, with cooked lentils at the bottom for contrast. The minus-fiber column is our arithmetic.

Food (USDA entry)	Carbohydrate (g)	Fiber (g)	Carbohydrate minus fiber (g)	Protein (g)
Firm tofu, raw, made with calcium sulfate	2.8	2.3	0.5	17.3
Cauliflower, raw	5.0	2.0	3.0	1.9
Tempeh	7.6	not listed	none	20.3
Avocado, raw	8.5	6.7	1.8	2.0
Hemp seeds, hulled	8.7	4.0	4.7	31.6
Almonds	21.6	12.5	9.1	21.2
Lentils, cooked (contrast)	20.1	7.9	12.2	9.0

The table runs from least carbohydrate to most. Firm tofu heads it, with 2.8 g of carbohydrate and more than 17 g of protein, and our guide to whether tofu is plant-based covers the one kind that isn’t. USDA’s tempeh entry gives no fiber figure at all, so it gets no number in our subtraction.

Almonds look like a mistake until you read the fiber column. Per 100 g they hold slightly more total carbohydrate than cooked lentils, 21.6 g against 20.1 g, but 12.5 g of the almonds’ share is fiber. They also bring 579 calories per 100 g, and hulled hemp seeds 553 with 31.6 g of protein, the most here. Avocado brings 14.7 g of fat per 100 g, and 6.7 g of its 8.5 g of carbohydrate is fiber. The old version of this page also listed berries and fermented foods, and gave per-serving carb counts for meat alternatives and vegan cheese. Those are gone, because it cited nothing for any of them.

What I’d do, and what the package won’t tell you

Read the Nutrition Facts panel, not the front of the box. “Keto” and “low-carb” have no FDA definition and neither does “net carbs”, so any company can print them, and the American Diabetes Association recommends checking total carbohydrate instead. Our minus-fiber column is one version of the net-carb sum, and the ADA notes that companies calculate it differently.

My own line, and it is a judgement rather than a finding, is to aim for the trial’s diet and not a keto app’s target. Swap some of the bread and potatoes for tofu, tempeh, nuts, seeds and low-starch vegetables, and keep some beans or lentils, which the Dietary Guidelines list among protein foods for vegans. The same guidelines tell vegans to diversify their plant protein sources and to prioritize targeted supplementation. If you have diabetes, or take medication for blood sugar or blood pressure, set the carbohydrate level with your doctor or a registered dietitian before you cut.

The part I’m least sure about is whether anyone can eat a vegan diet under 10% carbohydrate for years and do well on it. None of the studies cited here tested one, and the only vegan low-carb trial we read landed near 40%. If pasta is the carb you’d miss most, our low-carb vegan pasta roundup reads the labels box by box.

Sources

Gregory Knox

Gregory has been vegan since 2012 and came to it as a sceptic. He believed a balanced diet needed meat and dairy, so he read the studies and the books until he was sure it didn't. He still reads them, and turns what he finds into plain answers on food, ingredients, supplements and health questions. He also runs the technical side of Vegan Pots. Not a doctor, just very thorough.