Artificial Sweeteners and Gut Bacteria: The Human Trials

The claim that artificial sweeteners “wreck your gut bacteria” gets repeated as if it were settled. It traces back to one 2014 paper that was mostly about mice, and the human trials that followed disagree with each other in ways worth understanding before deciding what a diet soda does. This post walks through those trials: what each gave people, for how long, and what “altered the microbiome” meant in the data.

Where the claim came from

In 2014 a group at the Weizmann Institute reported in Nature that mice given commonly used sweetener formulations — saccharin among them, in their drinking water — developed glucose intolerance. Antibiotic treatment abolished the effect, and transferring stool from sweetener-fed mice into germ-free mice carried it across, pointing at gut bacteria as the mechanism in those animals.

The human part of the paper was far smaller. Seven healthy volunteers, five men and two women aged 28 to 36, took saccharin at 5 mg per kilogram of body weight a day for six days. Four of the seven ended up with impaired glucose tolerance; the other three did not.

So the finding that launched a decade of headlines rested, on the human side, on seven people, one sweetener and six days. That does not make it wrong. It makes it a pilot.

The 120-person follow-up

The same group ran the randomised version eight years later, published in Cell in 2022. It assigned 120 healthy adults to six arms: sachets of saccharin, sucralose, aspartame or stevia for two weeks at doses below the acceptable daily intake, 20 people per sweetener; a control sachet containing the glucose vehicle the sweetener sachets are built on; or no supplement.

Three findings matter. First, each sweetener, taken as a group, shifted stool and oral microbiome composition and the plasma metabolome relative to controls. Second, only saccharin and sucralose significantly impaired glycaemic responses; aspartame and stevia moved the microbiome without a detectable effect on glucose handling. Third, the effect was not uniform: when microbiomes from the top and bottom responders in each group were transferred into germ-free mice, the mice largely reproduced the glycaemic pattern of their human donor.

“Altered” here meant a statistically detectable compositional change over two weeks — not a collapse in diversity, and not something everyone experienced. The authors’ own summary was that sweetener consumption may produce person-specific, microbiome-dependent changes in glucose responses — a long way from “sweeteners cause diabetes”.

The trials that found nothing

Between those two Weizmann papers, independent groups tested the same idea against a control condition and found little or nothing.

• Sucralose, one week, high dose. A randomised, double-blind trial in Chile gave 34 healthy men 780 mg of sucralose a day or placebo for seven days. Neither glycaemic control nor insulin resistance moved, and at the phylum level the microbiome held steady in both groups. The men whose insulin response rose over the week carried less Bacteroidetes and more Firmicutes — in the placebo group as much as the sucralose group, so that association was not the sweetener’s doing.
• Saccharin at the maximum intake, two weeks. A double-blind, placebo-controlled trial took on the 2014 result directly, giving 46 healthy adults capsules twice a day for two weeks: 400 mg of sodium saccharin, the maximum acceptable daily intake, in one of four arms (placebo, saccharin, a sweet-taste-receptor blocker, or both). Nothing moved — not the oral glucose tolerance test, not microbial diversity, not composition at any taxonomic rank, not faecal metabolites, not short-chain fatty acids. A parallel ten-week mouse study was negative too.
• Aspartame and sucralose at realistic doses. Seventeen healthy adults in Canada took aspartame (0.425 g a day, 14 percent of the acceptable intake) and sucralose (0.136 g a day, 20 percent) in two 14-day periods with a four-week washout between them. The most abundant bacterial families and genera, the overall community structure and faecal short-chain fatty acids all came through unchanged.
• Stevia, twelve weeks. A Manchester trial compared 14 healthy adults taking five drops of stevia twice a day for 12 weeks with 13 controls. Neither alpha diversity (variety within a person) nor beta diversity (differences between people) changed, and no clear taxa differences turned up — though a machine-learning classifier still separated stevia users from controls about 75 percent of the time, which the authors read as a subtle pattern rather than a large-scale shift.

Four independent trials, four sweeteners, doses from realistic up to the maximum acceptable intake, and no significant change in microbiome composition or diversity in any of them. Against a seven-person pilot, that is a lot of weight.

The trials that found something

Two sucralose trials point the other way, and they share a feature the negative trials lack: they ran longer.

A ten-week study in Mexico had 20 healthy young adults drink water and 20 take 48 mg of sucralose a day. The sucralose group showed altered Firmicutes abundance, a three-fold rise in Blautia coccoides, a fall in Lactobacillus acidophilus, and higher serum insulin and glucose area under the curve. It was small and open-label, with plain water rather than a blinded placebo as the comparator.

A stronger design followed in 2025: a randomised, triple-blind, placebo-controlled trial in healthy lean adults given 30 percent of sucralose’s acceptable daily intake for 30 days. Insulin sensitivity fell by 20.3 percent, microbial alpha diversity dropped, faecal butyrate decreased and proinflammatory markers rose.

Side by side, the sucralose trials suggest a pattern: one week at a high dose did nothing measurable, while four to ten weeks at doses within the acceptable intake shifted both bacteria and glucose handling. That is a handful of small trials, not an established dose-time relationship, but it is the best current reason to keep the long-term question open.

Why the results disagree

The reviews that have tried to reconcile these studies land in similar places. A 2019 review in Advances in Nutrition concluded that of the sweeteners it covered, only saccharin, sucralose and stevia change gut microbiota composition, while polyols such as xylitol and maltitol make it as far as the large bowel and raise bifidobacteria numbers in humans. A 2023 review in Nutrients found a dysbiotic effect in some human trials but no significant impact in many other randomised controlled trials, and pointed to differences between studies in participant numbers, dietary habits and lifestyle — all of which shape the baseline microbiome and how it responds.

Three other things help explain the split:

• Sweeteners are chemically unrelated. Saccharin, sucralose, aspartame and stevia are grouped by taste, not biology. No controlled trial so far has shown aspartame or stevia impairing glucose tolerance, even where they nudged composition.
• Response is individual. Responders and non-responders were the point of the 2022 paper. A trial averaging 20 people can miss an effect confined to five of them, and a trial of seven can be dominated by them.
• “Altered” is easy to detect and hard to interpret. Sequencing picks up compositional shifts that may or may not change what the bacteria do; the 2023 review’s blunt assessment was that the field still has no agreed set of outcomes or biomarkers for defining a real effect. What diversity measures do and don’t tell you is covered in the gut health guide.

Observational data on habitual users adds little alarm. A 2015 cross-sectional analysis of 31 US adults — seven aspartame consumers, seven acesulfame-K consumers — found no differences in bacterial abundance or predicted gene function by sweetener use, though overall diversity did differ between consumers and non-consumers. A 2025 analysis of 572 adults in the Cancer Prevention Study-3 found no association between low-calorie sweetener intake and blood antibodies to flagellin or lipopolysaccharide, biomarkers used as proxies for intestinal permeability. Snapshots like these cannot establish safety, but the popular claim that sweeteners make human guts “leaky” gets no support from them.

What this means for a diet soda

What the evidence can carry:

• An occasional diet soda: probably little. Controlled trials running one to two weeks, at doses up to the maximum acceptable intake, did not measurably shift the microbiome or glucose control in healthy adults.
• Daily intake for months: an open question, specifically for sucralose and saccharin. The two longer sucralose trials found dysbiosis, lower butyrate and worse insulin sensitivity; the 2022 trial found saccharin and sucralose impaired glycaemic responses, in some people more than others. Neither has been shown for aspartame or stevia.
• The comparison that matters is usually sugar. Every trial here set a sweetener against placebo, water or no supplement, never against the sugared drink it replaces.
• Fibre is the better-studied lever. Faecal butyrate fell in the 2025 sucralose trial; what feeds short-chain fatty acid production is covered in Gut Bacteria and Fiber: Closing the Fiber Gap. No trial here tested whether a probiotic offsets a sweetener effect; the strain-specific evidence that does exist is in Probiotics and Gut Health.

Anyone managing blood sugar, pregnant, or taking glucose-lowering medication should talk to a clinician before changing sweetener intake.

Bottom line

The “sweeteners destroy your microbiome” story began with seven people and a lot of mice. The 120-person trial that followed found real but modest, person-specific shifts, with impaired glycaemic responses for saccharin and sucralose only. Four independent controlled trials lasting one to twelve weeks found no significant change, while two longer sucralose trials found dysbiosis and worse insulin sensitivity. For an occasional diet drink the measurable effect is probably small; for daily, long-term use of those two the case is unproven but not closed.