Where the number came from

The source is a 2018 paper in the journal mSystems describing the American Gut Project, a large citizen-science study with roughly sixty authors. Participants paid to have their stool sampled and sequenced, filled out questionnaires, and contributed to an open data set. By May 2017 it held sequence data from 15,096 samples and 11,336 people.

Somewhere inside that paper is the comparison that became the slogan. Here is what it was.

"Thirty plants" was a checkbox

This is the most important thing to understand about the claim, and it is almost never mentioned.

Participants answered a survey question. The exact wording, from the paper's methods:

"In an average week, how many different plants do you eat? (e.g., if you consume a can of soup that contains carrots, potatoes, and onion, you can count this as 3 different plants; if you consume multigrain bread, each different grain counts as a plant. Include all fruits in the total)"

People selected a range. The researchers then wrote, also in the methods:

"Only samples from people who consumed fewer than 10 types of plants a week or more than 30 types of plants a week were considered."

They compared the two extreme answer options and excluded everything in between.

That means the study never tested whether twenty was different from thirty. It never tested whether the relationship was gradual or whether there was a step somewhere. There is no dose-response analysis because the middle of the range was deliberately removed from the comparison.

Thirty was a pre-set answer option on a questionnaire, not a threshold the study discovered. If the survey had offered "more than 25" as its top bucket, we would all be trying to eat 25 plants a week and it would rest on exactly the same data.

How many people this rests on

The cohort had 11,336 people. The plant comparisons did not use them.

The metabolome comparison used 41 people in the high-plant group and 44 in the low-plant group. The diversity comparison used 42 versus 43. So the widely repeated claim rests on roughly 85 people, drawn from a much larger cohort by selecting the two extremes of a self-reported survey answer.

On top of that, only 14.8 percent of participants completed the validated food frequency questionnaire. The plant count came from the single self-reported question above, not from a validated dietary assessment.

What the study found

Setting aside how it was constructed, the findings themselves were modest and carefully worded.

People in the high-plant group had higher bacterial diversity and higher chemical diversity in their samples. Certain bacteria associated with short-chain fatty acid production were more common, but the statistical measure of how well that separated the groups was an AUC of 68.5 percent. In plain terms, that is weak-to-modest discrimination. If you were handed a sample and asked which group it came from, that model would be right somewhat more often than a coin flip, though not reliably.

The most concrete finding was about antibiotic resistance genes. The paper states that people eating more than 30 types of plants a week had significantly reduced abundance of resistance genes for several antibiotic classes compared with those eating 10 or fewer. The authors immediately added that these results are difficult to interpret given the range of antibiotics participants had taken, the many confounding variables, and the self-reported nature of the data. They also said that more controlled studies would be needed.

One more finding cuts against dietary labels generally. When the researchers clustered people by their full dietary questionnaires, producing recognizable groups like "vegan," those dietary clusters did not significantly relate to microbiome configurations: the categories people use to describe their eating did not map onto their gut bacteria.

The paper's own warning

The authors were candid about what their cohort was. This is from the paper:

"Because the AGP is crowdsourced and self-selected, and subjects generally support the cost of sample processing, the population is unrepresentative in several important respects, including having a lower prevalence of smoking and obesity and having higher education and income and underrepresentation of Hispanic and African American communities; generalization of the results should therefore be treated with caution."

People who pay to have their stool sequenced are not a random sample of anyone. They are wealthier, better educated, less likely to smoke, less likely to be obese, and disproportionately white. Every one of those things is independently associated with health outcomes and with diet.

The paper makes no causal claim anywhere. It uses "associated with" throughout. It is a cross-sectional, observational, self-selected, self-reported study, and it was offered as a hypothesis generator, which is a perfectly respectable thing for a study to be.

Has anyone replicated it?

No independent replication of a thirty-plant threshold in a general population could be located.

Two later papers are sometimes presented as support and are not. A 2023 analysis in the Journal of Nutrition and a 2021 analysis in the American Journal of Clinical Nutrition are both re-analyses of the same American Gut cohort. Looking at the same data again is not independent confirmation.

There is one independent finding, and it points somewhere slightly different. A 2024 study using the Baltimore Longitudinal Study of Aging, with 705 participants, found that a healthful plant-based diet index was associated with greater bacterial diversity, while an unhealthful plant-based diet index was associated with lower diversity.

Plant quality tracked with diversity in both directions. Plant count was not the variable.

The two trials that tested it

Two randomized controlled trials have used thirty-plus plants as an intervention. Both are worth knowing about, and both missed their primary endpoint.

A 2025 crossover trial in people with stage 3 to 4 chronic kidney disease compared at least 30 unique plant foods a week against 15 or fewer, six weeks per arm, in 25 participants. The primary outcome was uremic toxins, and the result was that plasma and urinary levels "did not consistently decrease across the cohort." Secondary and exploratory measures were more encouraging: symptom burden including constipation improved, and the microbiome shifted in ways the researchers found interesting. But it was small and in a specific disease population, and it did not hit what it set out to measure.

A 2026 trial in the British Journal of Nutrition is larger and more directly relevant: 399 participants over six weeks, testing a whole-food plant blend containing more than 30 ingredients at 30 grams a day against a cracker control and against a probiotic.

More bacterial species changed than in the control group: 57 versus 14. But on the trial's primary measure, the published finding is: "There were no significant between-group differences in microbiome health ranks of significantly changing species." The benefits that were reported were self-reported symptoms and energy.

The author list is the scientific leadership of ZOE Ltd, a commercial personalized-nutrition company: its co-founder, its chief executive, its chief scientist and its U.S. medical director are all authors. The primary endpoint was scored using ZOE's own proprietary microbiome health metric. The product tested is a thirty-plus-ingredient plant blend of the kind the company sells.

This was a company-run trial of a company product, judged by the company's own scoring system, and it missed its primary endpoint. That is not a reason to dismiss it, but it is a reason not to cite it as independent support for anything.

The disclosure that ties it together

One more fact changes how the slogan should be read more than any other.

Timothy Spector is a co-author of the 2018 American Gut paper. He is also the principal populariser of "thirty plants a week" in books, media and podcasts. And he is a co-founder of ZOE, the company that ran the trial above and sells a thirty-plus plant blend.

The chain runs: an observational finding in a paper he co-authored, to a consumer slogan he popularised, to a commercial product his company sells, tested in a trial his company ran and funded, which did not meet its primary endpoint.

None of that means the advice is bad. Variety in the diet is a sound idea with plenty of support from other directions. But "thirty plants a week" should not be presented as a neutral scientific consensus finding, and anyone repeating it as though a study proved it is repeating something the study did not say.

One more inconvenient finding

To be thorough, the premise underneath the slogan (that eating more fiber and more plant variety increases gut bacterial diversity) has been tested directly in randomized trials, and the result is not what you would expect.

A 2018 analysis in the American Journal of Clinical Nutrition pooled 64 randomized controlled trials covering 2,099 people. Fiber supplementation increased Bifidobacterium and Lactobacillus, and both increases were real and consistent. Fecal butyrate increased marginally, with a confidence interval that touched zero.

There was no effect on alpha-diversity.

So the diversity story rests on cross-sectional correlations (people who eat more plants have more diverse microbiomes) rather than on trials showing that adding plants increases diversity. Those are different claims, and only one of them has been tested properly.

How to count honestly, if you like the idea anyway

You should like the idea anyway. Variety is a reasonable thing to aim for, it is easy to remember, and it gets people to buy vegetables they have never cooked. Just count in a way that means something.

The official counting rule is extremely permissive. Remember the survey wording: each grain in a multigrain bread counts as a separate plant. By that rule, a doughnut made with wheat flour, soy lecithin and corn syrup is three plants. A packet of instant soup might be five. You can reach thirty without eating a single vegetable.

So count things you ate a meaningful amount of. A reasonable personal rule: it counts if you ate at least a tablespoon of it, or if it was a visible, identifiable component of a dish. A teaspoon of dried oregano in a sauce is a trace. A handful of fresh parsley in a salad is a plant.

Count across categories, not within one. Thirty plants made up of twenty-eight different lettuces is not variety. A useful check is to aim for a spread across six groups (vegetables, fruits, whole grains, legumes, nuts and seeds, and herbs and spices) rather than a raw total.

Legumes are the group people miss. They are also the highest-fiber foods on the list by a wide margin. In the Dietary Guidelines for Americans 2025 to 2030, beans, peas and lentils have moved from the Vegetables group into the Protein Foods group. They are the same foods in a different column, and easy to drop by accident if you are only working your way down a vegetable list.

Weekly, not daily. The point of the weekly frame is that it accommodates a normal life. A bag of mixed frozen vegetables, a can of three-bean mix and a jar of mixed seeds are how people hit variety without shopping every day.

Do not turn it into a spreadsheet. The honest case for this idea is that it is a nudge toward variety, not a dose-response target. Someone tracking to hit exactly thirty is over-applying a number that came off a questionnaire.

What variety is good for

Set aside the microbiome argument for a moment, because there are better ones.

Different plants contain different fibers, and fibers behave very differently: some thicken, some ferment, some add bulk. Eating one plant repeatedly gets you one profile. This is the best-established reason for variety and it does not depend on any gut bacteria finding.

Different plants also contain different micronutrients and different polyphenols. Variety is a straightforward hedge against gaps.

And practically, variety in the diet is a decent proxy for a diet built on whole foods, because it is hard to eat thirty different plants a week on convenience food alone. The habit tends to drag other things along with it.

Those are all defensible reasons to aim for variety. None of them requires the number thirty, and none of them requires a microbiome study.

Where a supplement fits

It does not replace this. A fiber supplement provides one fiber, or a few. That has real value: it closes a gap in quantity reliably, and it is consistent on the weeks when cooking does not happen. Read the grams per serving: many products deliver two or three, which is a small contribution against a ten-gram gap.

But quantity and variety are different problems, and a supplement only addresses the first. Anyone selling a powder as a substitute for eating plants is making an argument the evidence does not support.