Stool vs saliva vs blood microbiome tests

A microbiome test is not one biological measurement, it is sequencing (or culturing) microbes from a particular body site. Stool assays sample the distal colon and rectal lumen, where most consumer gut reports are anchored. Saliva assays sample the oral cavity, a distinct community with partial overlap to the gut via swallowing and the oral–gut axis. Blood assays detect bacterial DNA fragments in plasma; whether that represents translocation from the gut, assay contamination, or transient bacteraemia is contested, and blood is not a standard clinical substitute for stool composition. Skin kits sample yet another habitat, stool does not diagnose skin flora; see Gut–skin axis.

If your question is butyrate production, methane constipation, or FODMAP fermentation, a stool panel is the only consumer matrix that even samples the relevant geography, and even stool does not map the small intestine.

For report routing: Reading your microbiome report.


What not to conclude

MatrixWeak conclusionMore accurate framing
SalivaLow diversity means colonic dysbiosisOral ecology ≠ colon; different pH, oxygen, and diet exposure
SalivaOral probiotic will fix bloatingSwallowed bacteria may not colonise colon; strain and survival matter
StoolNormal stool rules out SIBOStool is downstream; overgrowth can exist proximally with altered distal output
Blood bacterial DNAProves “leaky gut” dysbiosisSignal may be low-level, episodic, or technical; not a validated dysbiosis assay
Cross-matrix comparisonOral improved so gut improvedDifferent pipelines and biology, not paired endpoints
Any matrixOne sample = whole-GI portraitGeography and timing dominate interpretation

What each matrix measures

MatrixPrimary biology sampledTypical consumer outputMain blind spots
StoolColonic lumen, fecal transit mixBacterial taxa, diversity, some archaea, pathway scoresSmall intestine, mucosa-attached communities, rapid proximal fermentation
SalivaOral mucosa, teeth, gingival creviceOral taxa (Streptococcus, Prevotella, Neisseria, etc.)Colon; esophageal disease unless specialised
Blood (plasma DNA)Circulating bacterial DNA fragmentsOften research-only; some wellness panelsCannot localise source organ reliably on most kits
Mucosal biopsy (clinical, not home kit)Epithelium-adherent layerPathology + culture/16S in researchInvasive; not consumer

The Human Microbiome Project catalogued distinct community profiles by body habitat, skin, oral, stool, and vaginal sites cluster separately (HMP Consortium, 2012). “Healthy microbiome” without a site qualifier is meaningless.


Clinical vs consumer availability by matrix

MatrixConsumer direct-to-consumerClinical/research use
StoolWidest DTC offering (16S, shotgun)Calprotectin, pathogens, C. diff, FIT, different assays from sequencing
SalivaOral health and cosmetic kitsPeriodontal risk, caries research; some H. pylori antigen tests (not 16S)
BloodRare; often bundled in disputed wellness panelsResearch on sepsis, cancer, and translocation; not guideline-backed for gut health
Breath (not microbiome sequencing)Some home hydrogen devicesLactulose/glucose breath testing for carbohydrate malabsorption and IMO/SIBO protocols

Stool sequencing and stool inflammation tests (calprotectin) answer different questions, both can be “stool tests” on a lab menu.


When oral results matter for gut symptoms

Oral–gut links are real but narrower than marketing implies:

  • Periodontal disease associates with systemic inflammation markers in epidemiology, causality and intervention benefit for distant symptoms are not established from oral 16S alone.
  • Dysphagia, reflux, nausea may originate above the colon; oral or esophageal workup precedes stool interpretation.
  • Nitrate-reducing oral bacteria affect salivary nitrite and may interact with host physiology, mostly research context, not consumer report lines.
  • PPI use alters upper GI pH and may shift both oral and stool reads (Medications and microbiome).

A high Fusobacterium read in saliva does not tell you whether Faecalibacterium is low in the colon. Treat oral kits as oral reports.


Blood-based claims, evidence limits

Studies detect bacterial DNA in blood from healthy volunteers at low levels, but:

  • Concentrations are far below stool biomass, signal-to-noise and contamination control dominate.
  • Post-prandial lipoprotein fractions can carry LPS and microbial ligands in metabolic research (Cani et al., 2007), that is biochemistry of metabolic endotoxemia, not a 16S profile of the colon (Metabolic health and endotoxemia).
  • Blood “microbiome” panels that imply gut dysbiosis diagnosis outrun validation.

Stool sequencing does not measure serum LPS. Blood metagenomics and stool metagenomics are different assays answering different questions.


How to choose a matrix for your question

QuestionSensible first matrixPoor fit
Colonic butyrate producers, diversityStoolSaliva, blood
Bloating after FODMAP foodsStool + dietary trial; breath if malabsorption suspectedOral kit alone
Gum disease, halitosisSaliva / dental examStool
Weight, insulin resistanceClinical metabolic labs; stool only as research adjunctBlood bacterial DNA as diagnosis
Post-infectious IBSClinical history + stool calprotectin if indicatedAny single DTC panel “subtype”

Context if you're reading a report

Marketing uses "microbiome test" for all three matrices. Readers apply oral or blood results to colon-focused questions (SCFAs, constipation, FODMAP fermentation) where the assay never sampled that geography.

Stool panels (16S or shotgun) sample distal gut contents; saliva kits profile oral taxa; blood assays detect bacterial DNA fragments with contested clinical meaning. None samples the full small intestine routinely.

That saliva reflects colonic SCFA producers; that blood bacterial DNA equals gut dysbiosis; that oral dysbiosis scores diagnose small-intestinal overgrowth; or that matrices are interchangeable for tracking diet response.

Related on this site: Human Microbiome Project Consortium, 2012, Nature , Rinninella et al., 2019, Microorganisms , Arumugam et al., 2011, Nature