The Blood Type Diet: Where Is the Evidence?

Doctors can already use genetic information to choose certain medicines. Could they do the same with diets? People respond remarkably differently to identical meals, yet popular ways of tailoring food to our biology have not stood up to rigorous testing.

Imagine a thousand cars: half run on gasoline, half on diesel. You want to know which fuel performs best.

So you randomly fill half the tanks with gasoline and the other half with diesel, without looking under the hood.

Roughly half the cars get the wrong fuel. When you compare the two groups, gasoline and diesel may come out looking equally bad.

First find out what engine you have. Then choose the fuel.

Could the same diet help one person, do little for another, and even harm a third? The promise of personalized nutrition is that we might know in advance, rather than spend years finding out through trial and error.

Human bodies, of course, are not divided into two incompatible models of engine. Most foods do not work in such stark either-or terms. But biological differences are real, and medicine already knows how to use some of them.

When Personalized Treatment Really Works

Consider cystic fibrosis, an inherited condition affecting the lungs and other organs. It is caused by changes in the CFTR gene. Different mutations have different consequences, and the particular variant can help determine which treatment will work.

In 2011, researchers tested ivacaftor in 161 patients carrying the G551D mutation. After 24 weeks, the improvement in lung function was 10.6 percentage points greater than with placebo. [1]

The drug addresses a specific defect in the protein associated with certain genetic variants. Other mutations call for different approaches.

Breast cancer offers another example. Some tumors make too much of a protein called HER2. Drugs such as trastuzumab were developed to target those tumors. The choice of treatment depends on the tumor’s biology, not simply its size or location. [2]

In several diseases, precision medicine has become part of routine care. Genetic tests and molecular features can sometimes point doctors toward a more effective treatment.

Researchers are now trying to apply the same principle to food. Instead of giving everyone general dietary advice, they seek to tailor recommendations to genes, metabolism, and the gut microbiome. Companies already sell programs built on that promise.

The idea is worth testing. Give different people the same meal, and their bodies may respond quite differently.

The Same Meal, Different Responses

In 2020, the PREDICT 1 team published findings from experiments involving 1,002 adults in the United Kingdom, including twins. [3]

Participants ate standardized meals while researchers tracked changes in blood glucose, insulin, and triglycerides afterward.

The coefficient of variation measures the standard deviation relative to the mean. These figures do not show how much a blood marker rose after eating. Glucose and insulin responses were assessed over two hours; the increase in triglycerides was measured over six.

Some of the spread reflects fluctuations within the same person. Meal timing, sleep, and physical activity can all matter. Yet repeated tests also revealed patterns characteristic of particular people and kinds of meals.

Genetics accounted for only part of the variation. The researchers also found associations with other characteristics, including the gut microbiome.

Suppose a particular meal sends your blood glucose higher than it does for most other people. What should you do? Cut out carbohydrates? Eat more fat? Serve yourself less?

That measurement cannot tell you. A brief rise in glucose does not, by itself, predict which diet will lead to a healthier weight, better cholesterol levels, or better health a decade from now.

To make personalized nutrition useful, we need to know which differences between people can genuinely help predict which diet will benefit them.

Do Personalized Diet Programs Actually Help?

In 2024, Nature Medicine published a trial involving 347 adults ages 41 to 70. [4]

One group received standard advice on healthy eating. The other followed a personalized program that took account of responses to food, the foods themselves, gut microbiome data, and medical history.

After 18 weeks, triglyceride levels in the personalized group were, on average, 0.13 mmol/L lower than in the control group, a statistically significant difference. The researchers found no convincing advantage for LDL cholesterol.

The personalized nutrition program was a commercial offering from ZOE. Its cofounders, employees, and consultants were among the study authors; some had financial interests in the company.

Those ties do not invalidate the findings. But the trial compared two complete programs, which also differed in how participants were advised and supported. It cannot separate the effect of biological testing from the effects of guidance, feedback, and encouragement.

In 2026, a review combined 15 randomized trials involving 2,351 people with overweight or obesity. [5]

Personalized programs led to slightly greater losses of body weight and fat mass than the comparison interventions. Across the trials, however, there was no statistically significant improvement in glucose measures or blood lipids. Confidence in the evidence ranged from moderate to very low.

So personalized programs sometimes outperform standard advice. Whether the biological matching itself accounts for the difference is another question.

One famous attempt to match diets to biology predates today’s genetic testing and microbiome analysis by decades.

Peter D’Adamo and the Blood Type Diet

In 1996, Peter D’Adamo published Eat Right 4 Your Type. It became an international bestseller.

The proposition was simple: your blood type tells you which foods suit you.

For type O, D’Adamo recommends more animal protein and fewer grains. For type A, he favors a largely vegetarian diet. Types B and AB have their own prescribed menus. [6] The Rh factor plays no part in D’Adamo’s recommendations.

A blood type is easy to check. No elaborate genetic testing, no continuous glucose monitor. If D’Adamo were right, it would be a remarkably convenient way to personalize a diet.

But why should a blood type determine this particular list of foods to eat and foods to avoid? That is the step in D’Adamo’s argument I would question first.

D’Adamo also offers an evolutionary story. Type O, he argues, belonged to ancient hunters; type A arrived with farming. But the genetic history is much older. Ségurel and colleagues showed that the lineages behind A and B have persisted in primates for tens of millions of years, long before agriculture or modern humans. Variants of O arose more than once over evolutionary time. The familiar tale of hunters giving way to farmers cannot explain the history of ABO. [7]

None of this makes blood type biologically irrelevant. ABO antigens are associated with blood clotting. People with type O have, on average, about 25% lower levels of von Willebrand factor than people with other blood types. That difference helps explain variations in the risk of thrombosis. [8]

Scientists have also reported associations between ABO genetic variants, LDL cholesterol, and several cardiovascular measures. An association is not a dietary prescription. [9] In a genetic study involving 8,956 people in Germany, ABO and the FUT2 gene were linked to features of the gut microbiota. [10]

D’Adamo’s proposed mechanism involves dietary lectins, which can bind to carbohydrate structures on cell surfaces. But the existence of those interactions does not show that everyday foods harm people with particular blood types, or that eliminating the foods he names improves health. Findings about clotting or the microbiome do not directly test his advice.

In a 2014 study of 1,455 people, some dietary patterns were indeed associated with more favorable metabolic markers. The apparent benefits, however, did not convincingly depend on whether the diet matched the person’s blood type. [11]

A 2018 analysis, with 973 participants at baseline, likewise found no additional advantage to a blood-type match. [12]

In 2021, researchers examined how 68 people of different blood types responded to a low-fat vegan diet. They found no convincing differences by blood type in the metabolic changes they measured. [13]

Taken together, these studies offer no convincing reason to choose a diet by blood type. They also weaken D’Adamo’s central claim: that matching food to ABO type brings an additional benefit.

Most of these studies did not randomly assign diets within each blood group. They mainly examined links between people’s usual eating patterns and measures of health. A randomized experiment would put the claim to a more direct test.

How Would You Test the Blood Type Diet?

The car example suggests a better experiment: test each fuel in each type of engine, rather than compare the average results alone.

The comparison helps explain the study design; it does not suggest that people with different blood types need mutually incompatible diets.

Imagine recruiting volunteers with types A and O, then randomly assigning people within each blood group to one of two nutritionally adequate but substantially different diets.

If the plant-based diet produces better results in both groups, it may be preferable. But knowing someone’s blood type has contributed nothing to the choice.

If people with type A benefit more from the plant-based diet, while those with type O do better on the alternative, that would support the case for matching diets to blood type.

The researchers would first have to decide which health outcome matters most. They would also need enough participants to detect a difference large enough to matter in practice, followed by an independent replication.

This is a proposed experimental design, not a description of a completed trial. Its purpose would be to test whether matching a diet to ABO type provides added benefit.

Existing studies have not found a convincing matching effect, making a large benefit less plausible. They cannot, however, tell us precisely how small a genuine effect might be.

Perhaps starting glucose levels, insulin sensitivity, or other metabolic traits would offer a better basis for choosing.

Can Genes Choose Between Low-Fat and Low-Carb Diets?

In 2023, the POINTS trial tested a more elaborate genetic idea. [14]

Using ten genetic variants, the researchers tried to predict who would lose more weight on a higher-carbohydrate diet and who would fare better on a higher-fat one.

The trial randomly assigned 145 adults with overweight or obesity to one of the two diets. For some, the assigned diet matched the genetic prediction; for others, it did not.

After 12 weeks, the main analysis of 122 participants gave the following results:

The adjusted difference was −0.6 kg (95% confidence interval, −2.1 to +0.9 kg; p = 0.50).

The genetically matched diet showed no statistically convincing advantage.

The larger DIETFITS trial followed 609 adults for a year. Participants were randomly assigned to a healthy low-fat or healthy low-carbohydrate diet. A profile based on three genetic variants did not show a statistically significant interaction with diet type for weight loss (p = 0.20). [15]

The researchers also asked whether a measure of insulin secretion at baseline could predict who would do better. They measured insulin 30 minutes after participants drank a glucose solution. Again, the interaction with diet type was not statistically significant (p = 0.47).

Earlier, I suggested that metabolic traits might prove more useful than blood type. DIETFITS makes me rethink that suggestion. Neither the genetic profile nor the metabolic measure tested here successfully identified which diet would work better.

There is an important distinction: insulin secretion is not the same as the body’s sensitivity to insulin. DIETFITS tested the former directly, not the latter. Still, its result makes it harder to argue that one blood test will reliably settle the choice between fat and carbohydrates.

I would take a different possibility more seriously now: for weight loss, lasting differences in the benefits of these two diets may be fairly small. Being able to follow a diet for the long haul may matter more than trying to pick the winner from a single biomarker.

That is my interpretation, not a conclusion of the trials themselves. Their results leave room for more sophisticated prediction models and for other diets. But after two independent trials, one lasting a full year, promises of reliably choosing between low-fat and low-carb diets from a gene profile or one blood test are hard to justify.

What Should You Eat Now?

If you want to lose weight, choose a nutritionally adequate way of eating that you can maintain. Look at weight trends measured under similar conditions, rather than day-to-day swings. If your concern is high cholesterol or impaired glucose regulation, follow the clinical measures and guidance relevant to that condition.

Try not to overhaul several habits at once when you change your diet. Otherwise, you may not know what made the difference. A single weight or glucose reading says little; the trend matters more. Think about nutritional adequacy, how well you tolerate the food, cost, and whether the plan is realistic for months at a time. If a medical condition requires a special diet, discuss major changes with your clinician.

If you already eat according to your blood type and feel well on that diet, the lack of evidence for the theory is not, by itself, a reason to stop. Check that the diet is nutritionally sound. Restricting grains and dairy, as D’Adamo recommends for type O, means paying attention to fiber and calcium. A strictly plant-based diet for type A requires adequate protein and a dependable source of vitamin B12. Feeling better, or seeing improved blood markers, might reflect the overall quality of your diet rather than a match with your blood type.

There is little reason to pay for genetic testing or a gut-microbiome analysis solely on the promise of finding your ideal ratio of fat to carbohydrate. Current studies have not shown that such tests reliably make that choice better.

The U.S. National Institutes of Health has launched Nutrition for Precision Health, a program investigating how to predict individual responses to food. [16] Research of this kind still needs to establish which biological differences are stable enough to guide dietary advice, and whether acting on those predictions improves health.

An engine determines what fuel a car needs. We have no comparably reliable way to classify people by the diets they need. For now, the diet that suits you is one you can stick with and that actually improves the health measures you care about.

References

1. Ramsey BW et al. (2011). A CFTR Potentiator in Patients with Cystic Fibrosis and the G551D Mutation. New England Journal of Medicine. https://doi.org/10.1056/NEJMoa1105185
2. National Cancer Institute. HER2’s Genetic Link to Breast Cancer Spurs Development of New Treatments. https://www.cancer.gov/research/progress/discovery/her2
3. Berry SE et al. (2020). Human postprandial responses to food and potential for precision nutrition. Nature Medicine. https://doi.org/10.1038/s41591-020-0934-0
4. Bermingham KM et al. (2024). Effects of a personalized nutrition program on cardiometabolic health: a randomized controlled trial. Nature Medicine. https://doi.org/10.1038/s41591-024-02951-6
5. Tran QD, Nguyen DK, Dang DK, Nguyen THH, Nguyen VK, Trinh TT. (2026). Effects of personalized nutrition on cardiometabolic biomarkers in adults with overweight or obesity: a systematic review and meta-analysis of randomized controlled trials. Nutrition & Metabolism. https://doi.org/10.1186/s12986-026-01130-3
6. D’Adamo PJ, Whitney C. Eat Right 4 Your Type. Penguin Random House. https://www.penguinrandomhouseretail.com/book/?isbn=9780399584169
7. Ségurel L, Thompson EE, Flutre T et al. (2012). The ABO blood group is a trans-species polymorphism in primates. Proceedings of the National Academy of Sciences, 109(45), 18493–18498. https://doi.org/10.1073/pnas.1210603109
8. Ward SE, O’Sullivan JM, O’Donnell JS. (2020). The relationship between ABO blood group, von Willebrand factor, and primary hemostasis. Blood, 136(25), 2864–2874. https://doi.org/10.1182/blood.2020005843
9. Li S, Schooling CM. (2020). A phenome-wide association study of ABO blood groups. BMC Medicine, 18, 334. https://doi.org/10.1186/s12916-020-01795-4
10. Rühlemann MC, Hermes BM, Bang C et al. (2021). Genome-wide association study in 8,956 German individuals identifies influence of ABO histo-blood groups on gut microbiome. Nature Genetics, 53, 147–155. https://doi.org/10.1038/s41588-020-00747-1
11. Wang J, García-Bailo B, Nielsen DE, El-Sohemy A. (2014). ABO genotype, ‘blood-type’ diet and cardiometabolic risk factors. PLOS ONE. https://doi.org/10.1371/journal.pone.0084749
12. Wang J et al. (2018). ABO Genotype Does Not Modify the Association between the “Blood-Type” Diet and Biomarkers of Cardiometabolic Disease in Overweight Adults. Journal of Nutrition. https://doi.org/10.1093/jn/nxx074
13. Barnard ND et al. (2021). Blood Type Is Not Associated with Changes in Cardiometabolic Outcomes in Response to a Plant-Based Dietary Intervention. Journal of the Academy of Nutrition and Dietetics. https://doi.org/10.1016/j.jand.2020.08.079
14. Höchsmann C, Yang S, Ordovás JM, Dorling JL, Champagne CM, Apolzan JW, Greenway FL, Cardel MI, Foster GD, Martin CK. (2023). The Personalized Nutrition Study (POINTS): evaluation of a genetically informed weight loss approach, a Randomized Clinical Trial. Nature Communications, 14, 6321. https://doi.org/10.1038/s41467-023-41969-1
15. Gardner CD, Trepanowski JF, Del Gobbo LC, Hauser ME, Rigdon J, Ioannidis JPA, Desai M, King AC. (2018). Effect of Low-Fat vs Low-Carbohydrate Diet on 12-Month Weight Loss in Overweight Adults and the Association With Genotype Pattern or Insulin Secretion: The DIETFITS Randomized Clinical Trial. JAMA, 319(7), 667–679. https://doi.org/10.1001/jama.2018.0245
16. NIH Common Fund. Nutrition for Precision Health. https://commonfund.nih.gov/nutritionforprecisionhealth