New Study Challenges Common Assumptions About the Carnivore Diet and the Gut Microbiome: What the Study Found and What It Didn’t

One of the most common warnings about the carnivore diet is simple: cut out every plant, and you starve the bacteria in your gut. Fiber feeds many of the microbes living in the colon, the reasoning goes, so a diet of only meat, fish, eggs, and dairy should leave the gut microbiome thinner, less diverse, and less healthy. It is a fair concern, and for years there was almost no long-term human data to test it.

A 2026 study in the journal Microbiota and Host set out to look (Karačić et al., 2026). The researchers compared the stool microbiomes of ten people who had followed a carnivore diet for around three years against those of 874 omnivores. The work drew attention because its headline result was not the collapse many would have predicted. But the picture is more nuanced than either “carnivore is proven safe for the gut” or “carnivore destroys it,” and how the study was built matters as much as what it found.

What did this study actually look at?

Before getting into what the numbers mean, here is the study in brief.

What is the gut microbiome, exactly?

The gut microbiome is the community of bacteria, viruses, and other microbes that live in your digestive tract, most of them in the large intestine. The microbiota is the collection of organisms itself; the microbiome usually refers to those organisms together with their genes and activity. A healthy adult carries hundreds of bacterial species down there, and scientists describe that community using a handful of standard measurements.

Alpha diversity describes the variety inside a single person’s gut, and it has two parts that often get blurred together. Species richness is simply how many different types are present. Evenness is how balanced their proportions are, whether a few species dominate or many share the space more equally. Beta diversity is a different idea: it measures how far apart two communities are from each other, which is how you tell whether two groups of people have genuinely different gut ecosystems. Microbial function refers to what those bacteria can actually do, the chemical reactions they are equipped to carry out.

One point matters before going further, because the rest of this article depends on it. A more diverse microbiome is often associated with good health, but diversity is not a health score on its own. High diversity is not automatically good, low diversity is not automatically bad, and the makeup and behavior of the community matter at least as much as the raw count of species.

Why is fiber usually considered so important?

Most dietary advice treats fiber as a cornerstone of gut health, and there is solid reasoning behind that. Humans cannot digest most plant fiber, so it travels to the colon largely intact, where certain bacteria ferment it. That fermentation produces short-chain fatty acids such as butyrate, acetate, and propionate, which feed the cells lining the colon and take part in metabolism and immune signaling (Facchin et al., 2024). Because particular fibers feed particular microbes, fiber-rich diets tend to support a range of fiber-fermenting species, and most public-health guidance encourages eating more of it.

A carnivore diet removes that fuel entirely. On paper, the expected result is fewer fiber-fermenting bacteria, less short-chain fatty acid production, and a less diverse community. That is exactly why a group of long-term, fiber-free eaters is scientifically interesting: they are a natural test of how much the gut microbiome truly depends on dietary fiber. None of this means every type of fiber behaves identically, or that diversity alone decides health, but it explains why removing all plant matter was expected to leave a clear mark.

How was the study done?

The study was cross-sectional, meaning it captured a single snapshot in time rather than following people as they changed their diet. The researchers recruited carnivore dieters through a Polish social-media figure with a large following in that community. To take part, people had to be adults who had eaten a carnivore diet for at least 12 months and who gave consent. The study deliberately excluded anyone who was pregnant, had a chronic inflammatory, metabolic, or autoimmune condition, reported digestive problems, took regular medication, or had recently used antibiotics or probiotics.

That left ten participants, six men and four women, with an average age of about 42. They had followed the diet for an average of 36 months, ranging from just under two years to five. On their questionnaires they described a healthy lifestyle overall, with roughly seven hours of sleep, regular exercise, and low stress, and they rated their own health markedly higher than the comparison group, the one background factor on which the two groups clearly differed. Their diets were not identical, though: all ten had cut out vegetables, legumes, and fast food, but some included dairy, fish, or poultry, and one relied on poultry rather than red meat.

For the microbiome analysis, each participant collected a single small stool sample with a swab, which was stabilized and shipped to one laboratory so that every sample was processed the same way. The lab used 16S rRNA gene sequencing, a widely used method that reads a specific bacterial gene to identify which broad groups of bacteria are present and in what relative amounts.

To estimate what those bacteria might be doing, the researchers ran a tool called PICRUSt2, which predicts a community’s likely functional capabilities by matching the bacteria it detects to reference genomes (Douglas et al., 2020). This detail is worth holding onto: the functional results are predictions inferred from which bacteria are present, not direct measurements of any chemical or activity in the participants. The team then used statistical models adjusted for age, body-mass index, sleep, and place of residence to test whether differences tracked with the diet rather than those other factors. The 874 omnivores came from the laboratory’s existing customer database and were selected to match the carnivores on those same characteristics.

What did the study find?

The two groups had clearly different gut microbiomes. But “different” is not the same as “worse,” and the details cut against a simple story in both directions.

Diversity and richness

The first surprise was in diversity. Three common measures that account for how evenly bacteria are distributed, the Shannon, inverse Simpson, and Pielou evenness indices, were slightly lower in the carnivore group, but the differences were not statistically significant. In plain terms, the meat-only eaters did not show the collapse in diversity that the fiber argument predicts. At the same time, one measure did differ significantly: Chao1 richness, an estimate of how many different bacterial types are present including rare ones, was higher in the carnivores (Karačić et al., 2026). So the accurate takeaway is not “diversity increased” or “diversity stayed the same,” but something more specific: the number of detectable bacterial types was actually higher, while the balance among them was, if anything, marginally less even, and not significantly so.

Microbial composition

Although overall diversity held up, the makeup of the two communities was distinctly different, and that difference was statistically strong. Among broad bacterial groups, the carnivores carried more of some normally rare phyla, including Synergistetes and Desulfobacterota, and less Firmicutes and Actinobacteria. Some of those fold-differences sound dramatic, with one rare group reported as roughly 90 times more abundant, but they need to be read in context: these are low-abundance organisms, so a large multiple still represents a small slice of the whole community. At the species level, the carnivore guts held more bacteria linked to protein fermentation and sulfate reduction, while several species generally regarded as beneficial, including Faecalibacterium prausnitzii and certain Bifidobacterium species, were less abundant (Karačić et al., 2026). Notably, some bacteria people often worry about, such as the Escherichia and Salmonella groups, did not differ significantly between the diets.

Predicted microbial functions

The largest differences were not in which bacteria were present but in what the software predicted they could do, and here a firm caution is needed. These functional results come from a computer model that infers metabolic capabilities from the bacteria detected; the study did not directly measure any of these chemicals or activities in the participants. With that in mind, the model flagged more than 300 predicted metabolic pathways and 13 functional modules as different between the groups, all of them higher in the carnivores.

The predicted differences pointed several ways at once. Some looked plausibly favorable: modules the researchers labeled as relating to gut-barrier function, and to the production of short-chain fatty acids such as acetate and propionate, were predicted to be more active. The model also predicted greater capacity for making certain B vitamins, including vitamin B12 and vitamin K, and for protein fermentation, which fits a high-meat diet. Others looked potentially concerning: modules the model associated with inflammation, cytotoxin production, methane-linked constipation, and sugar intolerances were also higher (Karačić et al., 2026). The researchers were careful to frame all of this as altered functional capacity rather than proven activity, and even suggested that the predicted rise in carbohydrate-handling capacity might reflect bacteria feeding on the gut’s own mucus lining rather than on dietary carbs (Tailford et al., 2015).

Two cautions are worth repeating. First, the statistical models explained only a modest share of the overall variation, at most about 17 percent for any single outcome, so the carnivore diet being the strongest measured predictor does not mean it explains most of what shapes these microbiomes. Second, a module named for inflammation or constipation is a label the model assigns to a set of genes; it is not evidence that the participants were inflamed or constipated. In fact, they reported regular bowel movements and good health.

What is genuinely surprising here?

The honest headline is narrow but real: in these ten people, removing all dietary fiber for years did not produce the sharp loss of microbial diversity the standard fiber argument predicts. Several species and pathways tied to fiber fermentation were preserved, and bacterial richness was actually higher, which challenges the simple assumption that no fiber must mean a depleted gut (Karačić et al., 2026). What it does not show is that fiber is unnecessary for human health, a far broader claim the study was never designed to test. Diversity and richness are not the same as health, and a microbiome can be diverse while still shifting in ways whose long-term consequences are unknown.

Why might diversity have held up without fiber? The researchers offer a few possibilities and present them as hypotheses, not conclusions. Gut bacteria may adapt to ferment animal-derived substrates such as protein and fat instead of plant fiber. Some may switch to feeding on host-derived glycans, the sugar chains in the mucus layer lining the gut (Tailford et al., 2015). And cutting out refined sugar and ultra-processed food, rather than the meat itself, might account for part of the effect. These are reasonable mechanisms, but none was directly demonstrated here, and each would need dedicated research to confirm.

Claim versus evidence

Because this study sits at the center of a noisy debate, it helps to line the loudest claims up against what it actually supports.

How does this fit with earlier research?

This paper does not stand alone, and reading it next to earlier work helps calibrate it. More than a decade ago, a well-known study showed that switching people to an entirely animal-based diet for just five days rapidly shifted their gut bacteria toward bile-tolerant types that ferment protein and fat (David et al., 2014). That established the short-term direction of change; the new study asks what a similar pattern looks like after years rather than days. The same research group had earlier published a single-person case study of one long-term carnivore (Karačić et al., 2024), and this work was designed to move beyond that one individual.

It is also worth separating this from the survey data often cited in carnivore discussions. A 2021 survey of 2,029 self-identified carnivore dieters reported high rates of subjective improvement (Lennerz et al., 2021), and we cover it in our breakdown of the Harvard-affiliated carnivore survey. But self-reported surveys cannot establish clinical outcomes, and this microbiome study does not change that. Finally, because the carnivore diet shares much of its metabolic profile with ketogenic eating, some of the overlap seen here echoes microbiome research on ketogenic diets (Lim et al., 2022); if you are weighing the two approaches, our comparison of carnivore versus keto goes deeper.

Why should researchers stay cautious?

The strongest reason for caution is the size and selection of the carnivore group. Ten people is a very small sample, and they were not chosen at random. They were recruited from a carnivore community, which means they were already committed to the diet and presumably doing well on it. Anyone who tried carnivore eating and felt worse or quit would never have appeared in the study, a pattern known as survivorship bias. The strict health criteria reinforced this, excluding people with digestive issues, chronic disease, or regular medication use, so the group represents healthy, long-term survivors rather than the average person who tries the diet. Their notably high self-rated health, the one factor that clearly separated them from the comparison group, fits that picture.

The comparison group brings its own caveats. The 874 omnivores are a useful baseline, and matching them to the carnivores on age, BMI, sleep, and residence strengthens the analysis. But they came from an existing commercial database rather than a study designed alongside the carnivore group, and a large comparison group does not offset a tiny treatment group. The evidence still rests on those ten carnivores. It is also worth noting that the laboratory which ran the sequencing both supplied the comparison data and sells microbiome testing, and two authors hold equity in it, which the paper discloses (Karačić et al., 2026).

Two conceptual limits matter most. Because the design was cross-sectional, with no measurement of these people before they started the diet and no follow-up over time, it can show association but not causation. And because the functional findings were predicted by software rather than measured, a predicted pathway is a hypothesis about capability, not proof of a biological or clinical effect. Ten carefully selected, healthy enthusiasts simply cannot tell us how the diet affects everyone who tries it.

What should you not take away from this?

It is easy to over-read a study like this, so it is worth being blunt. It does not prove that the carnivore diet improves gut health, that dietary fiber is unnecessary, or that the microbiome it described is clinically healthy, only that it is different and reasonably diverse. It does not establish that the diet is safe for everyone, that it prevents or treats any disease, or that the predicted inflammation- and cytotoxin-related modules caused any actual harm. And it does not establish long-term safety, which would require following people for years with real clinical measurements. The study is a useful early data point, not a verdict.

How should a reader interpret it?

The most reasonable reading is that the gut may be more adaptable than the simplest fiber argument allows, paired with real uncertainty about what those adaptations mean over decades. It is not a green light, and it is not a warning siren.

This is also where individual circumstances matter more than any single study. People with digestive disease, metabolic or kidney conditions, cardiovascular risk, or who are pregnant or taking medication were specifically excluded from this research, which means its reassuring-looking findings may not apply to them at all. Anyone in those groups, and really anyone considering a change this drastic, should talk it through with a doctor or dietitian who knows their history rather than leaning on a small study. If you want a grounded sense of what the diet involves day to day, our guide on what to eat on the carnivore diet is a practical starting point.

What research is needed next?

The authors are candid that this is a starting point, and the wish list for better evidence is clear. The most important step is simply more people, studied prospectively, meaning followed forward in time rather than captured in a single snapshot. Ideally that would include stool samples taken before someone starts the diet and repeatedly afterward, so changes can be tracked rather than inferred. Detailed food records would pin down what participants actually eat, since intake varied widely even in this small group.

On the laboratory side, two upgrades would help most. Shotgun metagenomic sequencing reads far more of the bacterial genome than the 16S method used here and identifies organisms and genes more precisely. And direct metabolomics, actually measuring the chemicals the bacteria produce, would replace today’s software predictions with real data. Adding blood biomarkers and clinical outcomes, comparing carnivore eating against several well-defined diets rather than one omnivore database, and having independent teams replicate the work would turn an intriguing observation into something firmer.

The bottom line

Taken together, this study offers useful early evidence that long-term carnivore eating can coincide with a gut ecosystem that is diverse and rich, yet substantially rewired in both its membership and its predicted chemistry. That is a more interesting result than either side of the carnivore debate usually allows. But it remains a small, cross-sectional snapshot of ten carefully chosen people, and its functional signals, some encouraging and some worrying, are predictions awaiting confirmation. The right response is not a louder claim in either direction, but better studies. For now, the gut looks more adaptable than the simplest fiber story suggests, and that is worth knowing without pretending it settles the question.