Key takeaways
- Sequencing results show each microbe as a share of the DNA the lab identified, so one group growing can make others look smaller even when their numbers never changed58.
- Healthy people’s stool can differ up to tenfold in total microbes, a gap that percentage-based reports can’t show7.
- Lab methods leave fingerprints: in a 2016 lab comparison, one common method missed Akkermansia entirely, and a database gap made another germ look like C. difficile5.
- In March 2025, the National Institute of Standards and Technology released a stool standard labs can use to check their methods, though nothing we found requires consumer labs to use it14.
- After a court struck down the FDA’s lab-test rule in 2025, a May 2026 House bill proposed a validity standard under federal lab law; it hadn’t advanced as of October 2026101113.
For six years, more than a dozen scientists at the National Institute of Standards and Technology (NIST) worked on an unusual product: human stool, blended, frozen and measured as precisely as they could manage1.
The federal measurement agency released it in March 2025. A unit comes as eight small frozen tubes, half from vegetarian donors and half from meat eaters. The data that come with it fill more than 25 pages, identifying over 150 kinds of microbes, plus more than 150 metabolites, the small molecules that living cells make and change1.
Why would anyone need a reference jar of stool? NIST molecular geneticist Scott Jackson put the problem plainly: “If you give two different laboratories the same stool sample for analysis, you’ll likely get strikingly different results”1.
Keep that in mind if a microbiome report is open on your screen. Our article on home microbiome tests covered why kits disagree and what to ask before you buy. This one starts after the results arrive.
Your report began as a scoop of stool and a pile of DNA fragments
The lab didn’t count living microbes. It pulled DNA out of your sample, read it in short pieces and matched those pieces against databases of known microbes5. Software then turned the matches into charts.
There are two main ways to read the DNA.
16S sequencing copies one gene, called 16S rRNA, over and over. Bacteria and archaea, another branch of single-celled life, carry this gene. Parts of it vary between groups, so it works like a barcode5. It costs less and suits large studies5.
The barcode has limits. In about 42% of bacterial genera, some members carry versions of this gene that are more than 97% alike, the usual cutoff labs use to lump sequences together5. In a 2016 comparison published in Frontiers in Microbiology, three common 16S software programs gave similar answers at the genus level, a group of related species. None did a satisfactory job of telling species apart5.
Shotgun metagenomic sequencing reads fragments of all the DNA in the sample. It can also pick up archaea, viruses and eukaryotes, the branch of life that includes yeasts and other fungi, and it often gets closer to the species5. It usually costs more, and its results depend on how many fragments are read and which genomes sit in the reference database5. In a 2021 study that used the chicken gut as a model, shotgun sequencing found rarer groups that 16S missed, as long as enough DNA was read6.
One more catch applies to both. DNA doesn’t tell you who’s alive. In a 2020 Dutch study of stool from 16 healthy volunteers, an average of about 39% of the microbial DNA came from dead or damaged cells or from loose DNA8.
The numbers are slices of a pie, not a head count
Sequencing results come as relative abundance: the share each microbe makes up of all the DNA the lab could identify5. A 2025 consensus of 69 experts from 18 countries, published in The Lancet Gastroenterology & Hepatology, says reports should present results that way, as percentages9.
Percentages always add up to 100. If one group grows, every other slice shrinks on paper, even if those microbes didn’t change at all8.
Picture a report where Bacteroides, a common gut group, drops from 30% to 20% between two tests. Maybe you have fewer Bacteroides. Or maybe they held steady while another group bloomed. The report can’t tell those stories apart.
Total numbers matter, too. When researchers at KU Leuven in Belgium counted microbial cells directly, they found up to tenfold differences in the microbial loads of healthy people7. Percentages alone can’t show how big a change was, or even which way it went7. A well-known pattern, in which Bacteroides and Prevotella seem to trade places, looked like an artifact of using percentages7.
A missing “good” microbe may be the method talking
Many reports single out microbes with good reputations, such as Akkermansia, Bifidobacterium and Lactobacillus. A low number can feel like a deficiency.
In the 2016 comparison, scientists ran a lab-made mix of known bacteria through 16S sequencing. The method didn’t detect Akkermansia or Clostridium at all, though together they made up about 5% of the mix5.
The barcode’s starting point, called a primer, matters as well. Primers that matched Clostridium and Lactobacillus well made those groups look bigger. Earlier work, the authors noted, found that one common primer pair undercounts Bifidobacterium5.
So does reading depth, the number of DNA fragments the lab reads. At 1,000 fragments, 16S found 48 groups; at 50,000, it found 72. Shotgun went from 58 to 1285.
The “number of species” on your report partly reflects how much the lab read. A popular diversity score, the Shannon index, barely moved5. Our diversity article explains those scores.
Software and reference databases have improved since 2016. Read these results as examples of how much method choices matter, not as today’s error rates.
Look for one more number. The 2025 consensus says reports should state how much of the DNA couldn’t be assigned to any known microbe9. A large unknown slice means the named groups describe only part of your gut.
A pathogen flag is a question for a clinician, not a diagnosis
The 2025 consensus says reports should list known disease-causing germs, such as C. difficile and Salmonella9. That statement drew 80.5% agreement, barely over the panel’s 80% bar and among the lowest in the document9.
Treat those flags with care. In the 2016 comparison, some Clostridium sordellii reads were labeled as C. difficile or as the germ that causes botulism, because the database had no genome for that species yet5.
A flagged germ’s DNA may also come from cells that are no longer alive8.
Infection testing is a different job. The same experts said multiplex PCR panels and lab cultures identify pathogens and don’t count as microbiome testing9. Our home-test article explains how clinicians test for C. difficile.
Don’t start antibiotics, herbal antimicrobials or a strict diet because of a flag alone. The panel agreed unanimously that only your referring physician or another health care professional should change your care based on results9.
Symptoms matter more than any flag. The National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) says adults with diarrhea should talk with a doctor right away if they have a high fever, black or bloody stools, severe pain in the belly or rectum, frequent vomiting or signs of dehydration14. Our guide to gut warning signs covers the rest.
Read your report, line by line
This guide is our own, built from the studies and consensus above.
| Report section | What it can tell you | What it can’t tell you | What to do |
|---|---|---|---|
| Method line (16S or shotgun, number of reads) | Which way the DNA was read, and how much5 | Whether the lab did it well | Compare only reports made the same way |
| “Unassigned” or “unknown” share | How much DNA matched nothing known9 | What those microbes are | If it’s missing, ask for it |
| Composition chart | The rough mix of major groups9 | Actual microbe numbers; species, on many 16S tests57 | Read the big groups; ignore small shifts |
| Diversity scores | How many kinds turned up and how evenly | Whether your gut is healthy | Treat as background, not a grade |
| “Beneficial bacteria low” | Fewer reads matched that group | Whether you lack it; methods can miss it5 | Don’t buy a supplement to fix a number |
| Pathogen flag | Some reads resembled a germ | Whether it’s alive or causing illness58 | Ask a clinician if you have symptoms |
| Gut score, dysbiosis index, Firmicutes-to-Bacteroidetes ratio | The company’s own formula | Your health; experts say leave these out9 | Ignore; see our home-test article |
| Predicted functions | Guesses based on microbe names | What your microbes are doing; experts advise against these9 | Ignore |
| “Healthy range” comparisons | Where you sit against the company’s reference group | Whether you’re unhealthy; strict ranges aren’t supported9 | Ignore |
| Food and supplement advice | The company’s suggestions | Proven benefit; experts discourage provider advice9 | Keep broad food advice; skip product pitches |
The panel discouraged test companies from giving any treatment advice, with 98% agreement9. Our probiotics article explains how to match a strain to a real goal if you’re still curious.
A frozen yardstick for labs arrived in 2025
NIST’s stool standard, Reference Material 8048, is meant for checking methods, comparing labs and qualifying instruments2. Its values are non-certified, meaning they’re meant for developing and aligning methods, and they’re valid through January 31, 20302. NIST’s project page explains the gap it fills: there had been no suitable standard for comparing one lab’s results with another’s4.
Even NIST’s own data show the limits of barcode sequencing. The material’s tables list the top 50 16S sequences in each stool pool, and many carry no species name2. In April 2026, 15 authors led by Jackson published a NIST technical report describing the material in detail3.
A lab can run the vials and see whether it gets NIST’s answer. If it doesn’t, the gap points to the lab’s method, not to anyone’s gut.
NIST sells the material through its online store4, and nothing we found requires consumer labs to use it. Ask your company whether it has checked its method against RM 8048 or another reference, and what it found.
The rules: a struck-down FDA rule and a bill that would raise the bar
Tests that a lab designs and runs itself are called laboratory-developed tests, or LDTs10. The FDA tried to oversee them as medical devices. In 2025, a court in American Clinical Laboratory Association v. FDA struck down that rule11.
That left the Clinical Laboratory Improvement Amendments (CLIA), the federal law on clinical labs, as the main authority1011.
On May 19, 2026, Rep. Neal Dunn of Florida introduced H.R. 8890, the Improving Clinical Laboratory Innovation and Access Act1013. It would amend CLIA to cover lab-developed tests and require tests for clinical use to have a “reasonable assurance of analytical and clinical validity”10.
Those two terms matter. Analytical validity means a test measures what it says, accurately and repeatably. Clinical validity means the result connects to a real health condition. For microbiome tests, that link is where the evidence is thinnest; experts reject strict healthy ranges and dysbiosis scores for lack of support9.
The bill would also create a public federal database of these tests, run by the Centers for Medicare & Medicaid Services, within two years10. A test that uses a sample people collect at home would still fall under these lab rules, and mailing out collection kits would count as part of running a lab10.
Many provisions wouldn’t start until two years after enactment, and the bill doesn’t mention consumer-ordered tests by name10. As of October 2026, GovInfo, the federal publishing office’s record, listed only the introduced version and no action after the May referral13. Check Congress.gov for its current status.
Meanwhile, hold companies to the expert panel’s standard: labs that are accredited, registered and regulated9.
Your stool sample is also a data file
The closest warning comes from a genetic testing company. In 2023, the Federal Trade Commission finalized an order against 1Health.io, formerly called Vitagene12.
The FTC said the company promised strong security but kept unencrypted health and genetic data in storage anyone online could reach12. It allegedly didn’t destroy DNA samples soon after analysis, as promised. In 2020, the FTC charged, it rewrote its privacy policy to widen the outside companies that could get customer data, and applied the change to people who had already tested12.
Under the order, the company had to pay $75,000, which the FTC planned to use for refunds. It also had to tell its contract labs to destroy DNA samples kept more than 180 days and to get clear consent before sharing health data12.
That case involved genetic tests, not microbiome kits, but the risk is similar. The 2025 consensus says testing providers should get consent people can withdraw, limit how long they keep data and remove identifying details9.
Before or after you test, check these:
- Sample: Is leftover stool destroyed, and when?
- Data: How long is your report kept, and is it tied to your name?
- Research: Is research use opt-in, or switched on by default?
- Deletion: Can you delete your account and data, and will you get written confirmation?
- Changes: Will the company ask before a new privacy policy applies to your old data?
How much weight each finding can bear
| Claim | Rating | Why |
|---|---|---|
| Lab methods shape what a report shows | Primer, depth and database effects are documented, and NIST built a standard because labs disagree156 | |
| A change in a percentage means that microbe changed | Percentages shift when other groups change, and loads vary up to tenfold78 | |
| A pathogen flag on a sequencing report means infection | Database errors and dead-cell DNA; experts leave infection testing to PCR and culture589 | |
| A supplement to raise a “low” microbe improves health | No accepted healthy range defines “low,” and experts discourage treatment advice from test providers9 | |
| A shared stool standard helps labs catch method problems | Built for method checks and described in 2026; we found no data yet on whether labs that use it agree more234 |
Natural-health traditions favor slow, steady change and close attention to your own body. A report that shifts because a lab changed a primer is a shaky guide for that kind of care. Your own record of meals, stools, sleep and energy is information the lab never sees.
If your report nudges you toward more beans, vegetables, whole grains or fermented foods, take the nudge. Those are changes our fiber, diversity and fermented foods articles support without any test. Skip the part that sells you a fix for a number.
Questions readers ask
Can I compare my new report with one from another company?
Not in any meaningful way. Different primers, reading depths and databases change the results56. If you retest, use the same company and method, and expect some movement anyway.
My 16S report says nothing about yeast or Candida. Did they miss it?
A 16S test can’t see fungi, because that gene is found in bacteria and archaea5. Shotgun sequencing can detect eukaryotes, the group that includes yeasts5, but experts say strict healthy ranges for microbes aren’t supported yet9. A worry about a yeast infection belongs with a clinician.
Should I stop my probiotic or medicine before a retest?
Not on your own. The 2025 experts agreed unanimously that people shouldn’t stop treatment or change their usual diet unless their physician recommends it9. Do write down everything you take, including supplements, since experts want that information recorded with the test9.
Can I get a second opinion on my raw data?
Experts support it with your written consent and with identifying details removed9. Ask the company whether it will release your data files. A clinician or dietitian who knows microbiome research can say what, if anything, deserves follow-up.
The bottom line
A microbiome report is a picture of DNA fragments, sorted by software and shown as shares of a pie. It can give you a rough sense of your gut’s major groups. It can’t count your microbes, confirm an infection or tell you which supplement you need, and many of its sections depend as much on the lab as on you.
Note the method, look for the unknown share, ignore single scores and product pitches, and take any pathogen worry to a clinician. Standards are improving, and the rules may tighten. Until they do, your plate and your own observations remain the most reliable tools you have.
Related reading
About this article
This article was researched using NIST records, peer-reviewed studies, an international expert consensus, federal bill records, NIH guidance and an FTC enforcement record, all listed below. It was reviewed in October 2026. It is general education and not medical, legal or financial advice.
Sources
- National Institute of Standards and Technology. “NIST Releases Reference Material to Aid Gut Microbiome Research.” NIST news release, March 17, 2025. nist.gov
- National Institute of Standards and Technology. “Reference Material 8048: Human Fecal Material.” Reference Material Information Sheet, March 7, 2025. tsapps.nist.gov
- Jackson SA, et al. “Characterization of Human Fecal Material (NIST RM 8048) from Omnivore and Vegetarian Donors.” NIST Special Publication 260-252, April 2026. nist.gov
- National Institute of Standards and Technology. “Human Gut Microbiome Reference Material.” NIST project page, updated May 2026. nist.gov
- Jovel J, Patterson J, Wang W, et al. “Characterization of the Gut Microbiome Using 16S or Shotgun Metagenomics.” Frontiers in Microbiology, 2016;7:459. frontiersin.org
- Durazzi F, Sala C, Castellani G, Manfreda G, Remondini D, De Cesare A. “Comparison Between 16S rRNA and Shotgun Sequencing Data for the Taxonomic Characterization of the Gut Microbiota.” Scientific Reports, 2021 (DOAJ record). pubmed.ncbi.nlm.nih.gov
- Vandeputte D, Kathagen G, D’hoe K, et al. “Quantitative Microbiome Profiling Links Gut Community Variation to Microbial Load.” Nature, 2017;551:507–511 (ProQuest record). pubmed.ncbi.nlm.nih.gov
- Galazzo G, van Best N, Benedikter BJ, et al. “How to Count Our Microbes? The Effect of Different Quantitative Microbiome Profiling Approaches.” Frontiers in Cellular and Infection Microbiology, 2020;10:403. frontiersin.org
- Porcari S, Mullish BH, Asnicar F, et al. “International Consensus Statement on Microbiome Testing in Clinical Practice.” The Lancet Gastroenterology & Hepatology, 2025 (Wageningen University repository). edepot.wur.nl
- U.S. House of Representatives. H.R. 8890, “Enhancing Clinical Laboratory Innovation and Access Act of 2026,” introduced May 19, 2026. GovInfo. govinfo.gov
- Balsano C. “Laboratory-Developed Tests…Continued.” Medical Laboratory Observer, July 7, 2026. mlo-online.com
- Federal Trade Commission. “FTC Finalizes Order with 1Health.io Over Charges it Failed to Protect Privacy and Security of DNA Data and Unfairly Changed its Privacy Policy.” FTC press release, September 7, 2023. ftc.gov
- U.S. Government Publishing Office. “H.R. 8890 (IH): Enhancing Clinical Laboratory Innovation and Access Act of 2026.” GovInfo bill details page, accessed October 2026. govinfo.gov
- National Institute of Diabetes and Digestive and Kidney Diseases. “Symptoms & Causes of Diarrhea.” NIH, last reviewed September 2024. niddk.nih.gov