Gut Microbiome Tests Disagree on the Same Stool Sample

Are Gut Microbiome Tests Worth It

Consumer gut microbiome tests are not worth buying to direct your diet or medical care. In the 2026 study by authors affiliated with the National Institute of Standards and Technology (NIST), seven consumer services tested the same stool material. They disagreed enough that differences between companies matched the differences between people. Other authors are at the University of Maryland and a consulting firm.

Across nine analyses from the commercial services and NIST, 1,208 unique taxa were identified. Only 3 genera appeared in every one of the nine profiles. Our arithmetic from the paper's reported numbers: 3 shared genera is about 0.2 percent of the 1,208 unique taxa found. After removing one outlier replicate from Company A, 17 genera were shared across all methods. That revised figure represents only 1.4 percent of the detected taxa. Seventeen out of eighteen monitored genera showed laboratory variability as large as biological differences between donors.

GI & Hepatology News quoted author Diane E. Hoffmann, who said "Healthy is not the same as typical." Hoffmann also said the tests may not be analytically valid, let alone clinically valid. When seven companies analyse the same stool and return different microbial profiles, you cannot use those profiles to decide what to change.

What the Study Compared

The 2026 benchmark evaluated seven direct-to-consumer microbiome services against homogenized human fecal material from a single donor. The research team ordered kits from each service, and the companies did not know about the study until their reports came back. NIST also tested the material as an eighth method alongside whole metagenome sequencing of seven additional donors. Every service received identical material, which ensured that any discrepancy originated from laboratory workflows rather than human biology.

The seven companies relied on next-generation sequencing methods to process the samples. Some services used 16S ribosomal RNA (rRNA) gene sequencing while others applied whole metagenome sequencing. The companies used different sample collection protocols, variable sequencing depths, and proprietary bioinformatic pipelines. Turnaround times ranged from two to eight weeks before consumer reports arrived.

Sequencing depth varied widely across the seven companies. Read depths stretched from under 20,000 reads to more than 10 million reads per sample. More reads did not reliably mean more species. For example, Company C reported over 600 species from 20,000 to 30,000 reads. In contrast, Company D reported 95, 98, and 99 species despite sequencing over 10 million reads. Reported genera per analysis ranged from 34 to 906.

Where the Reports Disagreed

Commercial reports generated from identical stool disagreed on which genera were present and how many. The authors tracked 18 genera that appeared in at least one replicate from every company to evaluate performance. Roseburia had the least agreement, with only 3 of the 9 methods inside the consensus range. Streptococcus was the only genus where all nine methods landed within the consensus range. No single method placed inside the consensus range for all 18 target genera.

Company A returned conflicting health classifications for identical stool material. Its first two replicates had just over 400,000 reads and shared 69 of the 70 and 71 genera they reported. Replicate 3 produced only 186,000 reads and had over 50 percent unidentified reads. Company A classed replicates 1 and 2 as healthy while rating replicate 3 as unhealthy. Seven of ten functional categories were below average for replicate 3, compared to one category for the first two replicates.

Divergence extended to a specific bacterial pathogen. Three companies detected Clostridioides difficile in the reference stool material. Four companies reported that the organism was absent. Reference populations also lacked common standards. One company measured against an internal average while another applied an interquartile range. Reference values for the five examined genera did not consistently overlap across companies.

Why the Same Stool Gives Different Reports

The companies differed in sample collection and handling, sequencing depth, data analysis and reference populations. The paper reports those differences but does not say how much each one contributed. The authors observed that conservation of genera between replicates within a single company ranged from 6 percent to 98 percent. Results within a single company were often consistent, but they could not be compared between companies.

What the Study Cannot Tell You

The benchmark trial measured analytical agreement rather than absolute diagnostic accuracy. Because the human fecal reference material lacks a defined ground-truth composition, the study cannot declare which company came closest to the true microbial makeup. The findings establish that the companies' results vary. They do not demonstrate that one specific testing workflow is correct.

The study protocol included several defined procedural limits. The research team evaluated only three replicates per company from one reference batch prepared from a single human donor. The material was sampled from a 50 mL conical tube rather than the provided devices or toilet paper. Samples were shipped at ambient temperature.

None of these findings indicate that the gut microbiome is irrelevant to human biology. Our guide on perimenopause bloating covers bloating and the estrogen-gut microbiome connection. Our review of fiber intake covers dietary fiber. The study shows that results from consumer testing services were poorly comparable.

How These Tests Are Regulated

Federal regulatory frameworks in the United States do not evaluate whether consumer microbiome tests provide clinically valid information. Under the Clinical Laboratory Improvement Amendments, known as CLIA, the Centers for Medicare & Medicaid Services oversees laboratory quality. CLIA rules require covered laboratories to verify analytical validity for their testing procedures. CLIA does not evaluate clinical validity.

No CLIA-approved proficiency testing programs exist for direct-to-consumer microbiome testing. CLIA also makes no distinction between facilities performing direct-to-consumer testing. In their legal evaluation, Hoffmann and colleagues observe that the Food and Drug Administration (FDA) has largely ignored consumer microbiome tests. The authors note the FDA likely views them as low-risk wellness tests.

The authors argue that many commercial microbiome marketing claims exceed wellness exemptions. Certain commercial services claim to restore function impaired by disease. The legal authors conclude that these commercial offerings lack analytical and clinical validity, fall outside low-risk wellness categories, and warrant more stringent federal oversight. Those conclusions represent the authors' legal analysis rather than an official FDA policy ruling.

Our Takeaway on Gut Microbiome Testing

Based on the 2026 study's results and our arithmetic, a gut microbiome test report is not a reliable guide to what you should eat or take. In the 2026 study, identical stool sent to seven testing services returned different lists of bacteria, and one company labelled replicates of the same stool both healthy and unhealthy. If you are an adult considering a commercial test out of curiosity, understand that the data reflects laboratory methods rather than definitive biology. The legal authors conclude the tests lack analytical and clinical validity.

This evaluation applies to healthy adults seeking actionable guidance on diet, supplements, or lifestyle changes. It does not apply to clinical diagnostic evaluations directed by a physician. Anyone experiencing gastrointestinal symptoms should consult a clinician. Standardized reference materials and minimum sequencing benchmarks would change this evaluation. The study authors call for minimum read depths, transparent analytical pipelines, and an industry consensus document to establish uniform testing baselines. Until independent standards exist and testing services demonstrate consistent outputs across shared specimens, commercial microbiome reports remain unvalidated decision tools. Before you pay for a gut microbiome test, ask a clinician which decision the result could change.