Keto is real medicine for epilepsy, but its edge for weight loss is small

Key takeaways

• In a 2020 Cochrane review, children with hard-to-treat epilepsy were far more likely to have seizures cut in half on a ketogenic diet, though the trials were few, small and unblinded²³.
• Across 13 trials lasting a year or more, keto diets beat low-fat diets by only about 2 pounds of weight loss, and LDL cholesterol rose slightly⁶.
• In a Stanford crossover trial, a Mediterranean diet lowered blood sugar about as well as keto, without keto’s rise in LDL or its nutrient gaps⁸.
• The FDA label for the drug dapagliflozin (Farxiga), revised in June 2026, lists a ketogenic diet as a trigger for ketoacidosis, a dangerous acid buildup that can strike when blood sugar is lower than doctors expect¹⁰.
• If you take medicine for diabetes or blood pressure, are pregnant, or have kidney disease, talk with a clinician before you start.

Charlie Abrahams was a little boy whose seizures medicine couldn’t control. His parents found an old diet described in a medical textbook and took him to Johns Hopkins Hospital in Baltimore to try it¹.

His seizures stopped within days, and he kept eating that way for five years, according to the foundation named for him¹. In 1997, his father, Jim Abrahams, produced the TV movie “First Do No Harm,” with Meryl Streep in the lead¹.

The diet was the ketogenic diet, developed in the early 1920s by Russell Wilder, a Mayo Clinic doctor¹. Several big American hospitals relied on it for epilepsy until new seizure drugs took over after World War II. By 1994, the foundation says, the diet had almost disappeared¹.

Today, keto is better known as a weight-loss plan. That leap, from hospital therapy to diet trend, is where the evidence thins out. We walk through what the best trials show, what can go wrong, and how to decide whether keto fits you.

Ketosis is a fuel switch, and it takes very few carbs to flip

When your body runs short of carbohydrate, it breaks down fat into ketone bodies, which can fuel cells, including brain cells, the National Cancer Institute (NCI) explains⁴. The main ketone in your blood is beta-hydroxybutyrate, or BHB. Living in this state is called ketosis⁴.

Getting there takes deep cuts. A major review of weight-loss trials defined a ketogenic diet as 50 grams of carbohydrate a day or less, or about 10% of calories⁶. Stanford’s Keto-Med trial aimed for 20 to 50 grams and cut out grains, beans, most fruit and added sugar to stay there⁸.

A blood ketone level above about 0.5 millimoles per liter is the usual marker of nutritional ketosis, according to a 2024 review in the journal Diabetes Therapy⁵.

Medical diets for epilepsy go further. In the classic version, fat outweighs protein and carbohydrate together by four to one, so about 90% of calories come from fat¹. Gentler versions, such as the modified Atkins diet, don’t restrict protein¹.

Ketosis is not ketoacidosis, a dangerous acid buildup in the blood that is a medical emergency. Certain medicines can push a person on keto toward it, as explained below¹⁰.

A century-old seizure therapy still earns its place

For children whose epilepsy doesn’t respond to medicine, the diet remains a real option. In 2020, Cochrane reviewers pooled 13 randomized trials with 932 people, 711 of them children². In a randomized trial, people are assigned by chance to one treatment or another.

In children, keto made it far more likely that seizures would drop by half or more². A summary in American Family Physician put it at about 453 children per 1,000 on the diet versus 78 per 1,000 on usual care³. For roughly every three children who tried the diet, one more reached that goal³.

Adults fared worse. No adult in the trials became seizure-free, the result for fewer seizures was uncertain, and adults were up to five times more likely to drop out².

Cochrane rated the evidence low to very low in certainty, mostly because the trials were small and no one could be blinded to what they ate². Even so, the reviewers called the diets a valid option for drug-resistant epilepsy when surgery isn’t suitable². The gentler modified Atkins diet may control seizures about as well; whether it also causes fewer side effects still needs study².

On our four-level scale, that makes the evidence promising but early. The rating reflects how few and small the trials are, not doubt about a therapy with a century of clinical use.

This isn’t a home project. In the United Kingdom, national guidance says children with drug-resistant seizures should be referred to a specialist and considered for the diet³.

For weight loss, keto’s edge over a year is about two pounds

The pitch is simple: cut carbs hard and the fat melts. Long trials tell a quieter story. In 2013, Brazilian researchers pooled 13 randomized trials with 1,577 adults who followed either a ketogenic diet or a low-fat diet for at least 12 months⁶. The keto groups lost only about 0.9 kilograms more, roughly 2 pounds⁶.

Keto did a bit better on triglycerides (a blood fat), HDL (“good”) cholesterol and diastolic blood pressure⁶. LDL cholesterol, the kind tied to clogged arteries, also rose slightly⁶. The authors called keto a possible tool against obesity but said the long-term differences looked small in clinical terms⁶.

A tightly controlled NIH study added a twist. In 2021, Kevin Hall’s team had 20 adults live at the NIH Clinical Center for four weeks, spending two weeks on each of two diets and eating as much as they wanted⁷.

One menu was plant-based and low in fat. The other was animal-based and ketogenic, with just 10% of calories from carbohydrate⁷. During the plant-based weeks, volunteers took in 550 to 700 fewer calories a day, yet rated their hunger and enjoyment of meals the same⁷.

Volunteers lost weight both ways, but body fat fell significantly only during the low-fat weeks, even though keto held insulin and blood sugar lower and steadier⁷. The NIH cautioned that the study wasn’t designed to guide weight-loss advice and that lab results may not hold at home⁷. Our companion article on low-carb diets weighs the bigger debate over carbs, insulin and weight.

In diabetes, a Mediterranean diet matched keto on blood sugar

Less carbohydrate means less sugar entering the blood, so keto appeals to many people with type 2 diabetes. The question is whether it beats a healthy diet that keeps beans, fruit and whole grains.

Stanford’s Keto-Med trial, published in 2022, tested exactly that⁸. Adults with prediabetes or type 2 diabetes spent 12 weeks on a well-planned ketogenic diet and 12 weeks on a Mediterranean-style diet, in random order. Both diets cut added sugar and refined grains⁸.

Among the 33 people with complete data, A1C, a blood test of longer-term blood sugar, fell 9% on keto and 7% on the Mediterranean plan⁸. That gap wasn’t statistically significant⁸.

Keto lowered triglycerides more, by 16% versus 5%⁸. But LDL rose 10% on keto and fell 5% on the Mediterranean diet⁸. Keto also supplied less fiber and less of nutrients including folate, vitamin C and magnesium⁸.

In the 12 weeks after the study, when people chose their own food, their eating looked more like the Mediterranean plan⁸. The researchers concluded that keto’s benefits weren’t large enough to justify giving up legumes, whole fruits and whole grains⁸. NIH grants and private gifts paid for the study, and the authors reported no conflicts⁸.

LDL cholesterol can climb on keto, sometimes sharply

On average, LDL creeps up in keto trials⁶⁸. Some people see far bigger jumps.

A 2024 study in the journal JACC: Advances recruited 80 adults who had eaten keto for an average of 4.7 years and whose LDL had reached 190 mg/dL or more, alongside high HDL and low triglycerides⁹. Their average LDL was about 272 mg/dL, and one person reached 591⁹.

Heart scans found no more plaque in this group than in 80 matched adults with much lower LDL⁹. But it was a one-time comparison of selected people, not a trial, and it can’t show what decades of very high LDL will do.

A practical step: get a cholesterol panel before you start and again a few months in. If your LDL climbs, ask about shifting your fats. Saturated fat raises LDL, according to genetic studies cited in the 2024 Diabetes Therapy review⁵, so olive oil, nuts, avocado and fish are better staples than butter and fatty meat. Our article on healthy fats covers those choices.

Medicines and conditions that change the safety picture

Some medicines and health conditions need a plan before day one.

SGLT2 inhibitors. These pills include dapagliflozin (Farxiga). The drug’s FDA label, revised in June 2026, lists a ketogenic diet among the triggers for ketoacidosis, along with eating fewer calories, illness, surgery and alcohol abuse¹⁰. Blood sugar may read lower than doctors expect, even under 250 mg/dL, so the danger can hide¹⁰.

Farxiga’s approved uses also include heart failure and chronic kidney disease, so check your medicine list even if you don’t have diabetes¹⁰. The label tells patients to stop the drug and get medical help right away if signs of ketoacidosis appear¹⁰.

Insulin and sulfonylureas. In the Keto-Med trial, people stopped sulfonylurea pills before the keto phase, and people on insulin, SGLT2 inhibitors or GLP-1 drugs couldn’t join⁸. Our article on medicines and wellness goals explains why these drugs can cause lows when carbs drop and how clinicians adjust them.

Pregnancy. The 2024 review found no evidence, including no randomized trials, supporting keto in pregnancy, and it flagged concerns about long-term ketosis and the baby’s developing nervous system⁵. Our low-carb article covers a birth-defects study tied to low carb intake.

Kidneys. Most trials also excluded people with kidney disease³. In the Cochrane review, extra calcium in the urine was more common on the classic diet than on modified Atkins². If you’ve had kidney stones or have kidney disease, ask first.

Type 1 diabetes. The review notes concerns about a higher risk of ketoacidosis and low blood sugar⁵.

Blood pressure medicines. Keto nudged diastolic pressure, the bottom number, down slightly in long trials⁶. If you take blood pressure pills, ask your clinician how to track your readings as your eating changes.

The “keto flu.” Some people feel rough at first. The 2024 review describes keto flu as mostly headache, brain fog and dizziness, sometimes with nausea and constipation⁵.

Eating disorders. If you have a history of disordered eating, a rule-heavy diet deserves a talk with your care team first. If restriction or weight loss is becoming a struggle, call the National Alliance for Eating Disorders helpline at 1-866-662-1235. If you’re in crisis, call or text 988.

Mental health, cancer and ketone drinks: early signals, not treatments

Some of the most hopeful keto research is in psychiatry. In 2024, Stanford researchers asked 21 adults with schizophrenia or bipolar disorder to eat keto for four months, with no change to their antipsychotic drugs¹¹. Each volunteer already had a metabolic problem, such as weight gain or insulin resistance¹¹.

Before the diet, 29% qualified as having metabolic syndrome, a cluster of risk factors for heart disease and diabetes. After four months, no one did¹¹. On average, people shed about a tenth of their weight, and a psychiatrist-rated measure of illness severity improved 31%¹¹.

That is a striking signal from 21 people, all on the diet, with no comparison group, so the diet can’t be separated from extra attention or time. The Stanford team called it a pilot to guide larger studies¹¹. Never stop psychiatric medicine to try a diet.

Cancer claims run further ahead of the evidence. NCI says there is no proof that popular diets, keto included, can slow or cure cancer or keep it from coming back, and it urges patients to talk with their care team before starting one¹².

Lab results are mixed. In a 2018 mouse study NCI described, keto made a class of targeted cancer drugs work better, but keto alone had uneven effects and seemed to speed up one leukemia¹³.

Exogenous ketones, meaning BHB drinks and pills, raise blood ketones without the diet. In 2022, NCI reported a mouse study in which BHB slowed colorectal tumors⁴. An NCI scientist called it a “don’t try this at home” case, noting that BHB is sold as a supplement but its risks and benefits need far more study⁴.

How strong is the evidence?

Claim	Evidence level	What the research shows
Keto cuts seizures in children with drug-resistant epilepsy, under a specialist team	Promising but early	Large benefit in four small, unblinded trials; Cochrane rates certainty low to very low23
Keto cuts seizures in adults with drug-resistant epilepsy	Not enough evidence	No adult became seizure-free; many dropped out2
Keto’s extra weight loss over low-fat diets is small	Strong evidence	About 2 pounds more over a year or longer, across 13 trials6
Keto lowers blood sugar in prediabetes or type 2 diabetes	Promising but early	A1C fell, but no more than on a Mediterranean diet in 12 weeks8
LDL cholesterol rises on average	Strong evidence	Seen in a 13-trial meta-analysis and a crossover trial; large in some people689
Keto eases serious mental illness	Promising but early	One 21-person pilot with no comparison group11
Keto treats cancer	Not enough evidence	Mixed mouse results; NCI finds no proof in people1213
BHB ketone supplements bring keto’s benefits	Not enough evidence	NCI says risks and benefits need far more study4
Keto is safe with an SGLT2 inhibitor and no medical plan	Evidence says no, or unsafe	FDA label lists keto as a ketoacidosis trigger10

Is keto a fit for you? A decision guide and a plate

Many keto eaters cook from scratch, read labels and have dropped soda, sweets and white flour. Those habits deserve credit, and they overlap with Keto-Med’s Mediterranean plan, which also cut added sugar and refined grains⁸. The supplement aisle is another matter, as the BHB findings above show⁴. Bring the skepticism you’d aim at a drug company to anyone selling a keto program.

This guide is our own, built from the sources above.

Your situation	Where you stand	First step
Healthy adult who enjoys the foods and wants to try it	Might try it	Get a cholesterol panel before and a few months in68
Any type of diabetes, or prediabetes	Talk with a clinician first	Ask about medicines, lows and ketoacidosis58
Take insulin, a sulfonylurea or an SGLT2 inhibitor	Talk with a clinician before day one	Medicines may need to change first810
High LDL, heart disease, kidney disease or past kidney stones	Talk with a clinician first	LDL can rise, and trials often excluded high cholesterol and kidney disease36
Child with epilepsy that drugs haven’t controlled	Only with a specialist team	Ask for referral to an epilepsy diet program3
Pregnant or trying to conceive	Avoid unless a specialist directs it	Safety in pregnancy hasn’t been shown5
In cancer treatment	Don’t use it in place of care	Raise any diet change with your oncology team12
History of an eating disorder	Talk with your care team first	Consider less rule-bound patterns

Questions readers ask

The bottom line

The ketogenic diet earned its place in medicine as a therapy for children whose seizures resist drugs, and it still holds that place. As a weight-loss plan, its edge over other diets is small.

For blood sugar, a Mediterranean diet matched it head to head and looked easier to keep. If keto appeals to you, build it from whole foods, check your LDL, and clear it first with a clinician if you take diabetes or blood pressure drugs, have kidney problems or are pregnant.

Related reading

• Some healthy habits clash with your medicines, so check these goals first
• Low-carb can mean 20 grams or 200, and what replaces the carbs matters most
• Fiber, form and a short walk shape how carbs affect your blood sugar
• Butter made the new food pyramid, but the evidence still leans toward plant oils
• Black-and-white thinking predicts regained weight, so learn to think in “mostly”

Sources

1. The Charlie Foundation for Ketogenic Therapies. “What is a Ketogenic Diet? Keto Diet Facts” (Keto Therapies). Charlie Foundation, accessed October 2026. charliefoundation.org
2. Martin-McGill KJ, Bresnahan R, Levy RG, Cooper PN. “Ketogenic diets for drug-resistant epilepsy.” Cochrane Database of Systematic Reviews, 2020 (abstract). cochranelibrary.com
3. Covey C. “Ketogenic Diets for Drug-Resistant Epilepsy” (Cochrane for Clinicians). American Family Physician, 2021. aafp.org
4. National Cancer Institute. “BHB May Help Prevent Colorectal Cancer.” Cancer Currents blog, June 2, 2022. cancer.gov
5. Firman CH, Mellor DD, Unwin D, et al. “Does a Ketogenic Diet Have a Place Within Diabetes Clinical Practice? Review of Current Evidence and Controversies.” Diabetes Therapy, 2024. link.springer.com
6. Bueno NB, de Melo ISV, de Oliveira SL, da Rocha Ataide T. “Very-low-carbohydrate ketogenic diet v. low-fat diet for long-term weight loss: a meta-analysis of randomised controlled trials.” British Journal of Nutrition, 2013. cambridge.org
7. National Institutes of Health. “NIH study compares low-fat, plant-based diet to low-carb, animal-based diet.” News release, January 21, 2021. nih.gov
8. Gardner CD, Landry MJ, Perelman D, et al. “Effect of a ketogenic diet versus Mediterranean diet on glycated hemoglobin in individuals with prediabetes and type 2 diabetes mellitus: The interventional Keto-Med randomized crossover trial.” American Journal of Clinical Nutrition, 2022 (Stanford-hosted PDF). med.stanford.edu
9. Budoff M, Manubolu VS, Kinninger A, et al. “Carbohydrate Restriction-Induced Elevations in LDL-Cholesterol and Atherosclerosis: The KETO Trial.” JACC: Advances, 2024 (Houston Methodist record). scholars.houstonmethodist.org
10. U.S. Food and Drug Administration. “FARXIGA (dapagliflozin) tablets, for oral use” (prescribing information). Drugs@FDA, revised June 2026. accessdata.fda.gov
11. Bai N. “Pilot study shows ketogenic diet improves severe mental illness” (on Sethi S, et al., Psychiatry Research, 2024). Stanford Medicine News Center, April 1, 2024. med.stanford.edu
12. National Cancer Institute. “Popular Diets, Supplements, and Cancer.” NCI, October 30, 2024. cancer.gov
13. National Cancer Institute. “Enhancing PI3K Inhibitor Response with Diet, Diabetes Drug.” Cancer Currents blog, August 8, 2018. cancer.gov