Can you do keto or low-carb as a vegan?

The objection

“Keto and low-carb are built on meat, eggs and butter. Take those away and you’re left with beans, lentils, bread and rice, which are exactly the foods keto bans, so a vegan version can’t work.”

The answer

Vegan low-carb eating has been tested in trials, and strict vegan keto is possible on paper, though we could find no trial of it and it takes careful planning. The objection is right that keto rules out the staples of most vegan cooking: Harvard’s review of the diet (first published 2018, last modified 2025) lists grains, starchy vegetables, most fruit, and legumes including beans, lentils and peanuts among the foods it excludes, and puts carbohydrate below 50 g a day, sometimes as low as 20 g, with 70 to 80% of calories coming from fat [1].

Many of the fats keto allows are plant ones, though: olive and coconut oil, avocado and coconut, nuts such as almonds, walnuts, pecans and macadamias, and seeds such as pumpkin, sesame, hemp and flax, alongside tofu and non-starchy vegetables like leafy greens, cauliflower, broccoli and mushrooms [1]. Keto keeps protein moderate, at 10 to 20% of calories, because the body can turn the amino acids in surplus protein into glucose and drop out of ketosis [1], so a vegan version leans hardest on those nuts, seeds, avocado and oils. Harvard warns that keto’s emphasis on saturated fat may have adverse effects on LDL cholesterol, and that the diet can be built instead on fats low in saturated fat such as olive oil, avocado, nuts and seeds [1]. The best-known vegan low-carb trial was not ketogenic and allowed more protein, taking more than half of it from wheat gluten, in seitan and a high-protein nut bread, and 23% from soy foods such as tofu and soy milk [2].

Has a vegan low-carb diet been tested?

Yes, in a handful of small trials, most of them allowing far more carbohydrate than keto. The Eco-Atkins trials tested a lower-carbohydrate vegan diet (about a quarter of calories from carbohydrate, well above keto levels) and found better cholesterol and slightly more weight loss than a high-carbohydrate vegetarian diet based on low-fat dairy and whole grains [2][3]. In Toronto, 47 overweight adults with raised cholesterol were randomly assigned either the vegan diet, prescribed as 26% of calories from carbohydrate, 31% from plant protein and 43% from fat, or a high-carbohydrate lacto-ovo vegetarian diet, both at 60% of their calorie needs [3]. After a month with all their food provided, both groups had lost about 4 kg, and LDL cholesterol had fallen about 8% more on the vegan diet (2009) [2][3]. Thirty-nine of them then carried on for six months choosing their own food. By the end, counting the first month, the low-carb group had lost 6.9 kg against 5.8 kg, a difference of 1.1 kg that only just reached statistical significance, and its LDL cholesterol had fallen 0.49 mmol/L more (2014) [2].

The limits matter as much as the results. Half of the low-carb group left the study before the end, against about a third of the comparison group, and their food diaries showed that people on the low-carb diet were really getting about 40% of their calories from carbohydrate [2]. The two diets also differed in more than carbohydrate: the comparison diet was built on low-fat dairy and whole grains and the vegan one on gluten, soy, nuts and vegetable oils, so the trials cannot say how much of the cholesterol advantage came from cutting carbohydrate and how much from the foods themselves [2][3]. The trial was part-funded by Solae, a soy company that employed three of its authors [2].

In shares of calories, the trial’s vegan diet sat a long way from typical keto even as prescribed, and further still as eaten, which is what the food diaries recorded over the six months [1][2].

Diet	Carbs	Protein	Fat
Typical keto [1]	5 to 10%	10 to 20%	70 to 80%
Eco-Atkins, prescribed [2]	26%	31%	43%
Eco-Atkins, eaten [2]	39.6%	22.7%	36.0%
High-carb, prescribed [2]	58%	16%	25%
High-carb, eaten [2]	53.9%	18.4%	27.5%

A second Toronto trial, of 164 people with type 2 diabetes over three months, with both diets recommended at 60% of calorie needs, found that a low-carbohydrate vegan diet and a vegetarian diet cut weight (5.9 kg against 5.23 kg) and HbA1c (0.99 against 0.88 percentage points) by similar amounts, with no significant difference between them (2022) [4]. A 2026 systematic review found seven randomised trials of mostly plant-based low-carbohydrate or ketogenic diets, 976 people in all, lasting 4 to 52 weeks. Weight fell by 5 to 7 kg and LDL cholesterol and HbA1c improved about as much as on the comparison diets or more, with moderate certainty for weight and cholesterol, but adherence fell once people were feeding themselves and the ketogenic versions needed supervision [5]. It is not wholly separate evidence, since it includes the 2022 Toronto trial and cites the 2009 and 2014 ones, as well as the pilot study below [5]. Strict vegan keto itself has not been trialled, as far as we could find. The nearest is a 45-day pilot study (2020) in which 48 people with obesity were given very-low-calorie ketogenic diets of 800 kcal a day or less, one group with its protein from plants, and all three groups lost weight and saw their total and LDL cholesterol fall [6].

The Toronto results came from diets built on soy, gluten, nuts and vegetable oils, and they may not carry over to keto in general, which tends to push LDL cholesterol up: a 2024 meta-analysis of 27 trials with 1,278 people found it 0.35 mmol/L higher after keto, and its authors urge caution [15]. How much it rises depends partly on body weight. In another 2024 meta-analysis of low-carb trials, LDL rose by 41 mg/dL where participants averaged a BMI under 25, did not change at 25 to 35, and fell by 7 mg/dL at 35 or more [19]. Coconut oil is one of the plant fats keto allows [1], and in a 2020 meta-analysis of 16 trials it raised LDL by 10.47 mg/dL compared with non-tropical vegetable oils [16], so a vegan keto diet heavy in coconut cannot borrow the Toronto trials’ cholesterol results.

Is low-carb healthier when the fat and protein come from plants?

Possibly: low-carb eating built on plant fat and protein has been linked to lower death rates than the meat-based kind, but the evidence is observational, and the newest and largest review (2026) found the link clear only in people who already had a chronic disease. In a 2018 pooled analysis of four cohort studies with 154,344 people and 30,959 deaths, diets that swapped carbohydrate for animal fat and protein were linked to an 18% higher death rate (hazard ratio 1.18), and diets that swapped it for plant fat and protein to an 18% lower one (0.82) [7]. A 2023 meta-analysis of ten cohort studies with 421,022 people found the plant-based low-carb pattern linked to a 13% lower risk of death from any cause (0.87), while the animal-based pattern’s 1.08 was not statistically significant, although it was linked to more deaths from cancer (1.16). Neither pattern was linked to deaths from cardiovascular disease, and the authors rated the certainty of their evidence low [8]. Results in that analysis varied widely from study to study [8]. The newest and largest of these reviews, of 18 cohort studies with 779,158 people and 219,457 deaths (2026), counted plant-based scores as healthy and animal-based ones as unhealthy. Among people who already had a chronic disease, the plant-based pattern was linked to a 28% lower death rate (0.72), evidence the authors rated high in certainty, but in the general population its 0.93 was not statistically significant, and the animal-based pattern was not linked to death rates in either group (1.04 in the general population, 1.05 with a chronic disease) [14].

A low-carb advocate would say, fairly, that none of these people were on keto. In the US cohort of the 2018 study, even the fifth of people eating the least carbohydrate got a median 37% of their calories from it, the authors note that few people ate a mainly plant-based low-carb diet, and they accept that some confounding factors may not have been adjusted for [7]. The same study found the lowest death rates among people getting 50 to 55% of their calories from carbohydrate, and in its pooled analysis higher death rates at both ends, below 40% and above 70% [7]. For weight, Harvard’s review concludes that after a year keto’s effects are not significantly different from those of conventional weight-loss diets, although the meta-analysis of 13 trials it cites found keto about 2 pounds ahead of low-fat diets at one year [1]. Diabetes UK says low-carb diets can be safe and effective in helping people with type 2 diabetes manage their weight, blood sugar and heart disease risk in the short term, but that there is no evidence they manage diabetes better than other approaches, including a healthy balanced diet, in the long term [9]. A 2024 meta-analysis of six trials lasting more than a year in people with type 2 diabetes found no significant difference in HbA1c, weight or LDL cholesterol between low-carb and comparison diets, though HDL cholesterol rose and triglycerides fell more on low-carb [17]. Keto also runs against the NHS’s advice for a healthy vegan diet, which is to base meals on starchy carbohydrates such as potatoes, bread, rice and pasta, and to eat beans and pulses [10].

What does vegan keto leave out?

Vegan keto cuts out pulses, whole grains and most fruit [1]. Harvard warns that keto can fall short on fibre, B vitamins, iron, magnesium and zinc, which are typically found in the whole grains it restricts, and suggests a registered dietitian’s help [1]. For vegans the squeeze is tighter, because the NHS’s vegan sources of iron are pulses, wholemeal bread, fortified breakfast cereals, dark green leafy vegetables, nuts and dried fruit, and strict keto rules out most of that list [1][10]. Vitamin B12 has to come from fortified foods or a supplement, as on any vegan diet, and the NHS also advises fortified foods or supplements for vitamin D, iodine, selenium and calcium [10]. Omega-3 is easier, since walnuts, hemp seeds and linseed (flax) appear on both the NHS’s list of vegan sources and Harvard’s list of keto foods [1][10]. In the first days or weeks people report hunger, fatigue, low mood, irritability, constipation, headaches and “brain fog”, and the suggested long-term risks include kidney stones, osteoporosis and raised uric acid [1].

Who should not try keto without medical advice?

Keto is not suitable in pregnancy, for children, or for people with kidney or liver disease without medical supervision, and anyone with type 1 diabetes or on insulin or blood-sugar tablets should speak to their doctor first [1][9][11]. Cutting carbohydrate while taking insulin or other diabetes medicines that can cause hypos raises the risk of a hypo, and anyone taking an SGLT2 inhibitor such as dapagliflozin should speak to the team that prescribes it first, as they may need to stop it: cutting carbohydrate on these drugs can raise the risk of diabetic ketoacidosis [9]. Dapagliflozin is also prescribed for heart failure and chronic kidney disease, and the NHS lists diabetic ketoacidosis among its serious side effects [18], so people taking it for those conditions should ask before starting keto too. The NHS tells people with type 2 diabetes not to change to a very strict, low-calorie diet without talking to a health professional first, as it may not be suitable for anyone taking insulin or other diabetes medicine, and not to stop a diabetes medicine unless a doctor or other health professional says so [12]. Speak to your GP before changing or stopping any medicine, including a statin, whatever a new diet does to your cholesterol.

Diabetes UK does not recommend low-carb diets for people with type 1 diabetes, and says they can affect children’s growth and should not be recommended for them [9]. Children with epilepsy that medicine has not controlled have been treated with a ketogenic diet since 1920 [1], but that is a medical treatment. Even a 2026 review that is broadly positive about keto lists acute pancreatitis, advanced liver or kidney disease and familial hypercholesterolaemia, an inherited form of high cholesterol, among the conditions the diet may make worse, and says it should be used in pregnancy only with extreme care and clinical supervision, and while breastfeeding only with extreme caution and regular checks on mother and baby [11]. It also names blood pressure tablets, diuretics, lithium and some epilepsy medicines among the drugs keto can interact with, so doses, blood levels or fluid balance may need checking, and lists porphyria and a few rare inherited metabolic disorders, such as those affecting fat breakdown, as reasons not to use the diet at all [11]. In a US case-control study, women whose carbohydrate intake before conceiving was in the lowest 5% got less than half as much folic acid from their food and were 30% more likely to have a baby with anencephaly or spina bifida (odds ratio 1.30, 2018), though a study of this kind can’t show that the diet was the cause [13]. The NHS advises anyone trying to get pregnant to take 400 micrograms of folic acid a day, ideally from three months before conceiving until 12 weeks into pregnancy, because it is hard to get enough from food alone, and some people, including those with diabetes, need a higher dose that a GP can prescribe [20].

Keto is also a diet of strict rules and banned foods. If cutting out foods or chasing weight loss starts to take over, a GP or an eating-disorder charity can help: Beat in the UK, Bodywhys in Ireland, ANAD in the US, NEDIC in Canada, the Butterfly Foundation in Australia, and EDANZ, which supports families and carers, in New Zealand.