At-home gut microbiome tests promise a picture of the bacteria in your gut, a "gut health score" and a personalized diet. In February 2026, scientists at the US National Institute of Standards and Technology (NIST), the agency that sets measurement standards, checked whether those reports can be trusted. Their answer, published in Communications Biology, is: not yet.
How the study worked
The team prepared one blended human stool sample, so that every portion was identical. They ordered three kits from each of seven direct-to-consumer services, 21 tests in all, filled every kit with the same material and sent them off like any customer. With the biology held constant, any difference between the reports could only come from the companies' methods.
What they found
- Very different lists of bacteria. The reports named 1,208 different bacterial groups in total, but only 3 genera appeared in every single result. Sequencing depth ranged from under 20,000 to over 10 million reads, and the number of species reported from under 100 to over 1,700.
- Differences as large as between people. For 17 of the 18 most common genera, the variation caused by the testing method was as large as, or larger than, the natural variation between different people.
- Disagreement on a pathogen. Three services reported Clostridioides difficile in the sample; four did not.
- "Healthy" and "unhealthy" for the same stool. One service labeled two of its three identical kits "healthy" and the third "unhealthy", with different diet advice. One kit with more than half of its reads unidentified still produced a full consumer report.
- No shared idea of "healthy". Genera often discussed in health reports, such as Bacteroides, Bifidobacterium, Faecalibacterium and Roseburia, varied widely, and each company compared them with its own reference ranges.
Why the results differ
Every step varies from one company to the next: how the sample is collected (a scoop of stool or a swab of used toilet paper), whether it is stabilized for the trip, how the DNA is extracted, the sequencing method (the 16S gene or the whole metagenome), how deep it goes, and the software and databases used to name the bacteria. On top of that, there is no agreed definition of a healthy microbiome. The authors call for reference materials and standards, the way clinical lab tests are standardized.
What it means for you
- If you already have a report, read it as a rough snapshot from one company's method, not as a measurement. Do not cut out foods or buy supplements because of a single score.
- If you are thinking of buying one, know that another company may tell you something different about the same stool.
- Digestive symptoms that last are a reason to see a doctor, who can order validated tests when needed.
What your DNA can add, and what it cannot
A DNA test reads your own genes, not your bacteria, so it cannot replace a microbiome test. But your genes do not change from one kit to the next, and a few of them shape the environment your bacteria live in: your secretor status (FUT2), lactase persistence and DAO. Large studies have linked lactase and FUT2 with gut bacteria such as Bifidobacterium. Our guide to microbiome genetics explains what they mean, and the free test reads them from the raw DNA file you already have.
Source: National Institute of Standards and Technology, Communications Biology, February 2026. The seven companies were not named in the study.