Diets for Type 2 Diabetes Compared: Keto, Mediterranean, Plant-Based, Low-Fat, DASH, and Intermittent Fasting

Quick answer: No single eating pattern has been shown to be best for everyone with type 2 diabetes. The American Diabetes Association’s 2019 nutrition consensus report, as reported by MDedge, said there is “no one correct eating pattern appropriate for all patients,” and ADA pages today list several patterns as options: Mediterranean-style, vegetarian or vegan, low-fat, very low-fat, low-carbohydrate, very low-carbohydrate, and DASH. The ADA says reduced carbohydrate intake has shown the most evidence for improving blood sugar, and that it can be used within different patterns. The research that compares diets is mostly short or medium term and not very consistent. So a practical way to read it is this: choose a pattern built on vegetables, whole foods, and fewer sugary drinks and refined grains, one you can keep doing, and work out the details, especially if you take diabetes medicine, with your care team. This page is general education, not a plan for you.

Why “the best diet for type 2 diabetes” is a trick question

People searching for the best diet for type 2 diabetes usually hope for a ranked list, with a winner at the top. The honest answer from the major diabetes organizations is that there is no single winner, and the reasons are worth understanding because they tell you how to use the rest of this page.

First, “best” depends on what you are measuring. A diet can be best for lowering HbA1c over three months, best for losing weight over a year, best for heart risk over a decade, or best for still being followed in two years. Those are different questions, and the answers do not always point to the same diet. Second, the studies do not line up neatly. They use different definitions (a “low-carbohydrate” diet in one trial may look very different from one in another), different comparison diets, different lengths, and different people. Third, people are different. Culture, budget, cooking skills, work schedules, other health conditions, and medicines all change what is workable and what is safe.

An ADA page on nutrition and wellness describes food choices as “personalized, equitable, shaped through the lens of culture and community.” Another ADA page says there “isn’t one eating plan that prevents or manages diabetes that anyone can use,” and that the ADA favors patterns you can sustain long term over fad diets. That is the idea this guide follows: lay out what each pattern is, what the evidence does and does not show, who needs extra care, and how to test what works for you.

The patterns people ask about, side by side

The table below summarizes the patterns the ADA lists, plus intermittent fasting, which people often group with diets even though it is about timing. The evidence column is deliberately short. It reflects what I could read, mostly through the ADA’s own pages and through summaries of the ADA’s 2019 consensus report, and it is not a full review of every trial.

Pattern	What it is	What the evidence summaries say	Cautions to raise with your care team
Mediterranean-style	Centers on plant foods, fish and seafood, and olive oil (ADA description).	In the PREDIMED trial, as summarized by the AMA, the group following this pattern had better blood sugar management after four years than the low-fat group and needed fewer glucose-lowering medicines. The AMA summary also calls overall results mixed. In a 2013 meta-analysis of long trials, it had the largest HbA1c effect of the diets compared.	Oil and nuts add calories. The ADA description says red meat is eaten rarely.
Vegetarian or vegan	Plant-based eating, with or without dairy and eggs.	Studies ran 12 to 74 weeks with mixed results. Both patterns lowered weight and lowered A1C by about 0.3 to 0.4 percentage points in people with type 2 diabetes (AMA summary of the ADA report).	Planning for vitamin B12 and protein matters, especially if you take metformin.
Low-fat and very low-fat	Cuts fat intake. The Ornish and Pritikin programs are very low-fat examples.	The Look AHEAD trial used a calorie-restricted low-fat plan with meal replacements and had moderate success. Lowering fat did not consistently improve blood sugar, and the benefit appears mostly tied to weight loss (AMA summary).	Very low-fat programs can be intensive. Pritikin was studied during a 26-day inpatient stay.
Low-carbohydrate and very low-carbohydrate	Reduces carbohydrate. Definitions vary from study to study.	The ADA says reduced carbohydrate has shown the most evidence for improving glycemia. In the AMA summary, one meta-analysis found a greater A1C benefit at 3 and 6 months, but not at 12 or 24 months. Long-term heart effects are unknown.	Medicines that lower blood sugar may need review. Very low-carbohydrate plans are the ones dietitians in one small survey most often advised against.
DASH	Vegetables, fruit, low-fat dairy, whole grains, poultry, fish, and nuts, with less saturated fat, red meat, sweets, and sugary drinks.	One small study in type 2 diabetes found better A1C, blood pressure, and cholesterol, with weight loss. A second found lower blood pressure but no change in A1C when combined with more activity (AMA summary). Evidence is limited.	Sodium and potassium advice may change with kidney disease or some blood pressure medicines.
Intermittent fasting	A schedule of when to eat, not what to eat.	The 2019 report’s summary, as relayed by the AMA, noted three studies showing weight loss without A1C improvement. Newer trials are described in our fasting guide.	Low blood sugar with insulin or sulfonylureas is the main concern.

The AMA article is a December 2019 summary of the ADA consensus report. The “survey” is a 2024 survey of US dietitians and nutrition counsellors, described by Dr. Alan Barclay of the University of Sydney’s Glycemic Index site. The article does not state the sample size, and it reflects clinician opinion, not patient outcomes.

Two other patterns appear in some reports. The ADA’s 2019 consensus, as reported by MDedge, also discussed Paleo and the USDA Dietary Guidelines. For Paleo, the AMA summary says very few studies exist, with 13 to 29 participants and usually three months or less, so the evidence is thin. The ADA pages I read today list seven patterns, without Paleo, which is a reminder that the lists differ slightly depending on which document you read and how it groups the patterns.

Dedicated guides for each pattern

What the research comparing diets actually found

The most useful single piece of context is how the big comparisons were done, so I will describe the two I could read about and then the limits that apply to all of them.

The 2013 meta-analysis. Ajala and colleagues pooled long randomized trials of dietary approaches for type 2 diabetes. As described in a physician commentary, it covered 20 randomized controlled trials with 3,073 participants, included only trials lasting longer than six months, and searched studies published through July 2011. It compared low-carbohydrate, vegetarian, vegan, low-glycemic-index, high-fiber, Mediterranean, and high-protein diets with control diets such as low-fat, high-glycemic-index, ADA, EASD, or low-protein diets. According to that description, low-carbohydrate, low-glycemic-index, Mediterranean, and high-protein diets each improved HbA1c more than their control diets, with the Mediterranean diet showing the largest effect. Low-carbohydrate and Mediterranean diets also produced more weight loss than their controls. In one trial, a vegan diet beat the ADA low-fat diet for HbA1c. I could not open the paper itself because PubMed was blocked, and the commentary did not report exact HbA1c or weight figures, so I am giving only the direction and ranking of effects.

The commentary also raised the limits. The control diets varied a lot, studies ran from six months to four years, and what counted as “low-carbohydrate” differed between studies. Its author argued that pooling such different studies may be problematic, and the authors acknowledged these limitations. That is a good reminder that a “ranking” from pooled trials is a rough guide, not a verdict.

The 2019 ADA consensus report. The report, published online in Diabetes Care on April 18, 2019 (Evert and colleagues) and reported by MDedge on April 26, 2019, did not crown a winner. It outlined the evidence for several patterns, and said that for preventing prediabetes or type 2 diabetes, the strongest research supports Mediterranean-style, low-fat, and low-carbohydrate patterns. Until comparative evidence improves, it recommended focusing on shared features: limiting sugars and refined grains, emphasizing non-starchy vegetables, and choosing whole foods over processed ones. It also said no evidence supports an ideal macronutrient split, so the split should be individualized, and that all adults with diabetes should be referred to individualized, diabetes-focused medical nutrition therapy. The MDedge page I read reported no A1C or weight numbers for any pattern.

What limits all of this. Several limits show up across the sources. Trials are usually short, often six months or less, and the benefits of some patterns, like low-carbohydrate eating, appeared at three and six months but not at 12 or 24 months in the meta-analysis the AMA described. People in trials are often motivated and supported, which may not match daily life. Very different diets were compared with very different control diets. And most trials did not follow people long enough to say anything strong about heart attacks, strokes, or kidney disease. I did not find a source I could read in full that shows one pattern clearly better than the others for those outcomes, and I would be wary of a source that claims to.

What the major guidelines agree on

If the differences between patterns are uncertain, the overlap between them is not. Looking across the ADA, the AMA’s summary of it, a University of Sydney summary of the 2024 ADA Standards, and the American Heart Association’s 2021 dietary guidance for heart health, the same themes keep coming up.

Theme	What the sources say	Where it came from
Vegetables	Emphasize non-starchy vegetables, which the ADA describes as full of vitamins, minerals, fiber, and phytochemicals, and low in calories and carbohydrate.	ADA page; ADA 2019 consensus as reported by MDedge
Whole foods over processed	Choose whole foods, and minimally processed foods, over processed ones. The AHA says to limit ultra-processed foods.	MDedge; AHA 2021 statement as summarized by the ACC
Whole grains and fiber	Choose whole grains over refined grains. The Sydney summary lists whole fruits, legumes, whole grains, nuts, and seeds among emphasized foods.	AHA 2021; Barclay summary of ADA 2024
Fewer sugary drinks and refined carbs	Minimize sugar-sweetened beverages, refined grains and starches, and processed meats. The AHA also says to limit added sugars.	Barclay summary of ADA 2024; AHA 2021
Healthy fats and proteins	The AHA favors liquid plant oils, plant proteins like legumes and nuts, fish at least twice a week (not fried), and avoiding processed meats.	AHA 2021
Salt and alcohol	The AHA says to choose foods with little or no salt, and says non-drinkers should not start drinking.	AHA 2021
Individualization	Food-based patterns should be adjusted to your preferences, culture, other conditions, and medicines.	ADA pages; MDedge

The AHA guidance is a scientific statement about cardiovascular health, not a diabetes guideline, so I use it only to show where heart and diabetes advice overlap. It lists white potatoes as an exception when it says to eat plenty of fruits and vegetables.

One more point from the ADA: the Diabetes Plate can act as a framework for all of the patterns above. The ADA page says it currently illustrates the low-carbohydrate pattern but its strategies can be adapted to any eating occasion. Our diabetes food guide collects single-food guides, so you can look up what specific foods hold, whichever pattern you choose.

Who needs extra care before changing how they eat

Changing a diet is often thought of as risk-free, but for some people it is not. The general principle across the sources is that if you take medicine for diabetes, a big change in what or when you eat can change how that medicine acts. I am not going to say what to do about any medicine. That has to come from the person who prescribes it.

Situation	Why it matters	Raise it with
Taking insulin or a sulfonylurea	Eating fewer carbohydrates or skipping meals can raise the risk of low blood sugar. The fasting commentary I read singles these out as needing careful planning.	Your prescriber, before you change how you eat
Taking other glucose-lowering medicine	Some medicines have their own cautions when intake changes sharply, such as with dehydration or very low carbohydrate intake. I could not verify specifics by drug, so ask.	Your prescriber or pharmacist
Kidney disease	Protein, potassium, and sodium advice can differ. I did not find a source I could read that settles this for every pattern.	Your kidney or primary care clinician and a dietitian
Pregnancy or breastfeeding	Nutrition needs and glucose targets are different, and people in this group are commonly told to avoid fasting.	Your obstetric and diabetes care team
History of eating disorders	Strict rules, counting, and fasting can be harmful for some people with that history.	A clinician with eating disorder experience
Heart disease or high blood pressure	Some patterns change sodium, saturated fat, and potassium intake.	Your clinician and a dietitian

This table is a general prompt for questions, not a complete list of risks, and it is not medication advice.

The ADA’s 2019 report said that all adults with diabetes should be referred to individualized, diabetes-focused medical nutrition therapy. In practice, that usually means a registered dietitian, and it is one of the best-supported things on this whole page. If you have not seen one, ask your care team for a referral, or ask whether your insurance covers the visit. I could not verify coverage rules for this page, so check with your plan.

How to choose a pattern you can actually follow

If the research cannot name a winner, the choice has to rest on fit. The sources point to sustainability, and that is a practical test you can use. Think through these questions before you pick, and write down your answers.

What do you already eat on a normal week? The easiest change is usually the one closest to your current food.
What does your culture or family eat? A pattern that asks you to give up every shared meal is hard to keep.
What is your food budget and your time to cook? Fresh fish, lots of nuts, and specialty products add up.
What medicines do you take? This can decide which patterns are safe to start without help.
What are you trying to change first? Blood sugar numbers, weight, blood pressure, or cholesterol may point to different starting points.
How do you do with rules? Some people like clear rules and others rebel against them.

If this sounds like you	A pattern to discuss	Why it might fit
You like a clear framework and want to focus on blood sugar	Lower-carbohydrate eating	The ADA says reduced carbohydrate has the most evidence for improving glycemia, though benefits in one meta-analysis faded by 12 months.
You enjoy cooking, vegetables, fish, and olive oil	Mediterranean-style	Evidence is broad, and the ADA lists it first among patterns.
You already eat little meat or want to eat less	Vegetarian or vegan	Studies show weight and modest A1C improvements, and need planning for B12.
You also have high blood pressure	DASH-style	The pattern is built around blood pressure, with limited diabetes-specific evidence.
You eat late at night or snack often	A time-based approach, if safe for you	Timing may help some people, but see our fasting guide for the safety limits.
You take insulin or a sulfonylurea	Any pattern, planned with your care team first	Medicine timing and doses matter more than the pattern.

These are starting points for a conversation, not matches that the research has proven.

Seeing what a pattern is doing to you

Population studies tell you what happened on average. Your own numbers tell you what is happening to you. If you have a glucose meter or a CGM, a diet change is a good time to look at them, but it takes a plan, or you end up chasing single readings.

A simple approach is to decide with your care team what you will track and for how long, then change one thing at a time. Our post-meal glucose tracker is a free tool for recording what you ate and what your numbers did afterward, and our guide to reading CGM data explains the patterns on a trace. Ask your care team which times and targets make sense for you, since they depend on your medicines and your goals. Keep in mind that one meal is a small sample. The same meal can look different from day to day because of sleep, stress, activity, and illness.

Also track more than glucose. Hunger, energy, sleep, digestion, cost, and whether you looked forward to the next meal all tell you whether you can keep this up, and those will decide whether any pattern works over a year.

A little arithmetic: what “low” and “moderate” carbohydrate mean in grams

Carbohydrate talk is easier with numbers. The arithmetic below uses 1,800 calories a day as an arbitrary round number to make the math clean. It is not a target and not advice, and your own needs will differ. Carbohydrate has about 4 calories per gram, so grams equal calories times the carbohydrate share, divided by 4. The shares are illustrations only. As the 2013 meta-analysis commentary pointed out, definitions of “low-carbohydrate” vary between studies, and I did not find a single official gram cutoff in the sources I read.

Illustrative carbohydrate share	Share of an 1,800 calorie day	Grams per day	Per meal if split in three
Very low share	10% of calories	about 45 g	about 15 g
Low share	26% of calories	about 117 g	about 39 g
Moderate share	45% of calories	about 202 g	about 67 g
Higher share	55% of calories	about 248 g	about 83 g

Computed as 1,800 × share ÷ 4, rounded. The shares are examples chosen for illustration, not official definitions or recommendations.

Now the same idea with real foods. Our white rice guide puts a cup of cooked white or brown rice at about 44 to 45 g of carbohydrate, and our banana guide puts a medium banana at roughly 27 g. A meal with one cup of rice and a banana comes to about 71 to 72 g. Cut the rice to half a cup and it is about 49.0 to 49.5 g. Half a cup of rice on its own is about 22.0 to 22.5 g. Against the per-meal figures above, the full cup and banana is already over the “moderate” per-meal figure of about 67 g, and a very-low-share day has room for only a small amount of either. The point is not that these foods are bad. It is that the same plate fits very differently depending on the pattern, which is why reading labels and counting what you eat is useful early on. Our nutrition label guide shows how.

Common mistakes when comparing diets

Treating one study as a verdict. A single trial, or a news headline about one, can point in any direction. Look for reviews and for what the major diabetes organizations say.
Copying someone else’s plan. What works for a person who takes no medicine can be unsafe for someone who does.
Changing medicines to fit a diet. Never change the dose or timing of a prescription because a plan says so. Plan it with your care team first.
Judging by one reading. One high or low number after a meal says little. Patterns over days say more.
Cutting out whole food groups with no replacement. If a pattern removes grains, dairy, or meat, make sure fiber, calcium, protein, and B12 are covered. Ask a dietitian.
Chasing perfection. The ADA favors patterns you can keep over fad diets. A plan you drop after three weeks is worth less than a milder one you keep.
Ignoring the rest of the picture. Sleep, activity, and stress all matter. Even a short walk after meals is part of the picture.

Myth	What the sources I read say
“One diet beats all the others for diabetes.”	The ADA consensus said there is no one correct pattern for all patients, and pooled trials were too different to rank firmly.
“Low-carbohydrate is proven best.”	The ADA says reduced carbohydrate has shown the most evidence for improving glycemia, but the AMA summary notes one meta-analysis found benefit at 3 and 6 months and not at 12 or 24 months, and long-term heart effects are unknown.
“Plant-based diets don’t work for blood sugar.”	Studies of vegetarian and vegan patterns reported modest A1C reductions, about 0.3 to 0.4 points, though results varied.
“Fasting is just another diet.”	It is a timing approach, not an eating plan, and low blood sugar with certain medicines is a real risk.
“If it’s on the diabetes diet list, it’s safe for me.”	Safety depends on your medicines, kidneys, heart, and history. Ask your care team.

Questions for your doctor or dietitian

Given my medicines, which eating patterns would you be comfortable with me trying, and which would you want to adjust first?
Could you refer me to a registered dietitian for diabetes nutrition therapy, and is it covered?
What numbers should I track, how often, and for how long?
What would make you want me to stop or change the plan?
How should I handle low blood sugar if I cut carbohydrates or skip meals?
Do my kidneys, heart, or blood pressure change what you would suggest?
If I want to try a lower-carbohydrate or plant-based pattern, what nutrients should we watch?
When should we check my A1C or labs to see whether this is working?

The complete guide to type 2 diabetes covers the other pieces of care this page leaves out.

The bottom line

There is no ranked winner among diets for type 2 diabetes, and the people who write the guidelines say so. Several patterns have some support, low-carbohydrate eating has the most evidence for improving blood sugar in the short term, the Mediterranean pattern has broad support, and the overlap among all of them (vegetables, whole foods, fewer sugary drinks and refined grains) is the part with the strongest agreement. The rest comes down to safety with your medicines, your preferences, and whether you can keep it going. Pick a starting point from the guides above, plan it with your care team, track a few things for a few weeks, and adjust.

Frequently asked questions

What is the best diet for type 2 diabetes?

The sources I read do not name one. The ADA’s 2019 consensus report, as reported by MDedge, said there is no one correct eating pattern for all patients, and ADA pages list several options. Reduced carbohydrate has shown the most evidence for improving blood sugar, while Mediterranean-style eating has broad support. The best fit depends on your medicines, preferences, and what you can keep doing.

Which diet lowers A1C the most?

It depends on the study. A 2013 meta-analysis of long trials, as described in a physician commentary, found low-carbohydrate, low-glycemic-index, Mediterranean, and high-protein diets each improved HbA1c more than their control diets, with the Mediterranean diet showing the largest effect. The studies were varied, so treat it as a rough guide.

Is low-carb better than Mediterranean for diabetes?

The sources do not settle this. The ADA says reduced carbohydrate has the most evidence for improving glycemia, and the AMA summary says one meta-analysis found a low-carbohydrate benefit at 3 and 6 months but not at 12 or 24. The Mediterranean pattern has support for broader outcomes. See the dedicated guides for each.

Can I eat a vegan diet with type 2 diabetes?

Studies of vegetarian and vegan patterns, summarized by the AMA, ran from 12 to 74 weeks with mixed results, and reported A1C drops of about 0.3 to 0.4 percentage points along with weight loss. Plan for vitamin B12 and protein, and ask your care team, especially if you take metformin.

Is keto safe for type 2 diabetes?

It depends on your medicines and health. In one small survey of US dietitians described by the Sydney Glycemic Index site, very low-carbohydrate or ketogenic diets were the pattern most often advised against, mainly for being restrictive and hard to sustain. Medicines that lower blood sugar may need review, so talk to your care team first.

Is intermittent fasting a diet?

It is a timing approach rather than an eating plan, and the 2019 report’s summary said three studies showed weight loss without A1C improvement. Newer trials are described in our fasting guide. Low blood sugar with insulin or sulfonylureas is the main safety concern.

What do all the diets have in common?

Emphasis on non-starchy vegetables, whole foods over processed ones, whole grains over refined, and fewer sugary drinks and refined carbohydrates. The ADA’s 2019 report suggested focusing on these shared features while comparative evidence improves.

Do I need a dietitian?

The ADA’s 2019 consensus said all adults with diabetes should be referred to individualized, diabetes-focused medical nutrition therapy, which is usually delivered by a registered dietitian. Ask your care team for a referral and check what your plan covers.

How long should I try a new diet before deciding?

The sources I read give no standard answer. Many trials ran three to six months, and some benefits faded later. A sensible approach is to agree with your care team on what to track, check in after a few weeks, and look again at three months.

Can I change my diet without changing my medicine?

Not always. If you take insulin or a sulfonylurea or another medicine that can lower blood sugar, a big change in carbohydrate or meal timing can change how it acts. Do not adjust any medicine yourself. Plan changes with your prescriber first.

Sources checked

Not read: NIDDK, Mayo Clinic, and Harvard Nutrition Source pages were not reachable from my tools, and PubMed and PubMed Central were blocked, so none of them are cited here.

About the author: Charles Kirkland is the founder of Metabolic Clarity Labs and author of The Journey Back. He is not a physician. This is general education, not a medically reviewed treatment plan.

Sources checked October 31, 2026. Discuss personal decisions with a qualified clinician. Do not start, stop, or change prescribed treatment based on this article.