Artificial Sweeteners: The Honest Take on Aspartame, Sucralose, and Stevia

Two Headlines, One Morning, Zero Contradiction

In July 2023 the internet got two press releases about aspartame within hours of each other, both carrying WHO branding, and they appeared to say opposite things. One announced that aspartame had been classified as possibly carcinogenic to humans. The other announced that the acceptable daily intake was unchanged. Comment sections concluded that the WHO was either covering something up or contradicting itself.

Neither. The two documents came from bodies with different jobs, and once you know what each job is, the split decision stops being confusing and becomes the single most useful thing in the whole file.

The International Agency for Research on Cancer runs hazard identification. Its question is whether a substance is capable of causing cancer under some set of circumstances, at some dose, in some organism. It does not ask how much exposure that would take, and it explicitly does not estimate risk at real-world intakes. That is why the IARC monograph list contains processed meat, working night shifts, and drinking beverages above 65 degrees Celsius alongside plutonium and tobacco. The list is sorted by strength of evidence that a thing can cause cancer, not by how much cancer it causes.

The Joint FAO/WHO Expert Committee on Food Additives runs risk assessment. Its question is the one an ordinary person is actually asking: given how much of this substance people consume, is there a problem? JECFA looked at the same literature IARC looked at, found no reason to move, and left the acceptable daily intake at 0 to 40 mg per kilogram of body weight.

The IARC classification rested on limited evidence for hepatocellular carcinoma in humans, limited evidence in experimental animals, and limited mechanistic evidence. In IARC's vocabulary, limited is a specific technical grade, not a synonym for weak-but-worrying. It means a positive association was observed in studies where chance, bias, and confounding could not be ruled out with reasonable confidence.

The US Food and Drug Administration then did something regulators rarely do in public, and said flatly that it disagreed. FDA scientists had reviewed the same studies in 2021 and identified what the agency called significant shortcomings in the work IARC relied on. The FDA's position is that aspartame remains safe under the approved conditions of use, and its acceptable daily intake of 50 mg per kilogram is unchanged.

What Group 2B Actually Classifies

Group 2B is IARC's third tier out of four, and it is the category for agents where the human evidence is limited and the animal evidence is less than sufficient. It is the default holding pen for substances that have been studied enough to raise a question and not enough to answer it.

The classification tells you nothing about potency and nothing about exposure. Two substances in the same IARC group can differ by many orders of magnitude in how much cancer they cause in the real world, because the group reflects the confidence of the evidence rather than the size of the effect. This is not a criticism of IARC, which says so itself in every monograph preamble. It is a criticism of how the classifications get reported.

There is a second thing worth knowing, and it is the reason the aspartame file is unusually thick. Aspartame is one of the most studied food additives in the human food supply. The FDA reviewed more than 100 studies before and during its approvals, covering reproductive effects, neurological effects, carcinogenicity, and metabolism, and it has monitored the literature continuously since the last approved use in 1996. Sucralose went through more than 110 studies. Neotame, more than 110. This is not a case where a novel compound slipped through on thin data.

None of that means the question is permanently closed. It means the burden sits with new evidence, and the new evidence that triggered the 2023 review was a handful of observational cohorts we will get to shortly.

The Acceptable Daily Intake, Converted Into Real Life

An acceptable daily intake is not a danger threshold. It is the amount considered safe to consume every single day across an entire lifetime, and it is constructed with a deliberate cushion. Regulators find the highest experimental dose that produced no observed adverse effect in animal testing, then divide by a safety factor, typically one hundred, to account for differences between species and differences in sensitivity between people. So the ADI is already roughly a hundredfold below the highest dose that did nothing detectable to a test animal.

The FDA also reports that for every sweetener it has approved, the estimated daily intake of the general population, including heavy consumers, comes in below the ADI. The ceiling is not something the population brushes against. It is a line drawn well out past where anyone stands.

Now do the arithmetic on yourself, because the abstraction is what makes people nervous and the concrete number is what dissolves it.

Aspartame, at three body weights

The FDA ADI for aspartame is 50 mg per kilogram of body weight per day. JECFA uses 40. A can of diet soda contains roughly 200 to 300 mg of aspartame depending on formulation, a figure WHO used in its own 2023 communication.

• A 70 kg (154 lb) lifter: 3,500 mg by the FDA limit, 2,800 mg by JECFA. That is roughly 12 to 17 cans a day at the high end of the aspartame range, or 14 to 17 at the low end. WHO's own framing was more than 9 to 14 cans daily for a 70 kg adult.
• A 90 kg (198 lb) lifter: 4,500 mg by the FDA limit, 3,600 mg by JECFA. Somewhere between roughly 12 and 22 cans a day.
• A 110 kg (242 lb) lifter: 5,500 mg by the FDA limit, 4,400 mg by JECFA. Roughly 15 to 27 cans a day.

Every one of those numbers describes a daily habit sustained for life. Not a bad weekend. Not the day you were bored at work and drank four Diet Cokes. A permanent intake, every day, forever, and even then you would only be touching a line that sits a hundredfold below the no-effect dose.

The packet version, from the FDA's own table

The FDA publishes a chart converting each ADI into tabletop sweetener packets for a 60 kg (132 lb) person, assuming one packet is as sweet as two teaspoons of sugar. It is the single most useful public document on this topic and almost nobody has seen it.

For a 60 kg person, the ADI works out to 75 packets of aspartame, 23 of sucralose, 23 of acesulfame potassium, 45 of saccharin, and 27 of rebaudioside A. Scale those linearly by body weight and a 90 kg lifter gets about 112 aspartame packets or roughly 34 sucralose packets a day.

Thirty-four packets of Splenda is a comically large amount of Splenda. It is also, notably, not infinite, and sucralose is the sweetener where a heavy user could conceivably get within sight of the number. Somebody who sweetens four coffees, two protein shakes, a bowl of oats, and a pre-workout, all with sucralose-heavy products, is not at 34 packets but is no longer in the trivially-safe zone that aspartame occupies. That is worth knowing, and it is an argument for variety rather than an argument for fear.

The Numbers Table

Two things in that table deserve a second look. First, an ADI has not been specified for monk fruit or for thaumatin, and that is not an oversight or a loophole. A regulator can decline to set a numeric limit when the evidence shows safety at levels far above the amount needed to sweeten anything, which makes a ceiling meaningless in practice.

Second, the low ADI numbers are not a danger ranking. Neotame's 0.3 mg per kilogram looks alarming next to aspartame's 50 until you notice that neotame is 7,000 to 13,000 times sweeter than sugar, so the amount needed to sweeten a drink is correspondingly microscopic. That is exactly why the packet column exists and why it lands in the same range as the others. Compare packets, not milligrams.

The Observational Studies That Keep Restarting the Scare

The modern round of worry does not come from the rodent work. It comes from large prospective cohorts, and the most influential of them is NutriNet-Sante.

Debras and colleagues followed 102,865 French adults from 2009 to 2021, accumulating 708,905 person-years and a median follow-up of 7.7 years, during which 3,358 cancers were diagnosed. Higher consumers of artificial sweeteners had a hazard ratio for overall cancer of 1.13 against non-consumers, with a confidence interval running from 1.03 to 1.25 and a trend test at P = 0.002. Aspartame specifically came in at 1.15 and acesulfame-K at 1.13. Breast cancer with aspartame was 1.22, confidence interval 1.01 to 1.48. No association appeared for prostate cancer.

Those are real numbers from a serious cohort and they should not be waved away. They should also be read for what they are. A hazard ratio of 1.13 is small. The confidence intervals nearly touch 1.00 in several of the headline results, meaning the data are compatible with effects close to nothing. Dietary intake was captured by repeated 24-hour records, which is better than a food frequency questionnaire and still an estimate. And the deepest problem is one no statistical adjustment fixes: people who drink a lot of diet soda are systematically different from people who do not.

That last point is called reverse causation. In most fields it is a footnote. In sweetener research it is the dominant explanation on the table. People reach for artificial sweeteners because they are carrying more body fat, because they have been told to cut sugar, because they have prediabetes or a family history of it. The diet soda is a marker of the condition it gets blamed for. Any cohort that finds diet soda drinkers have more metabolic disease has, at minimum, rediscovered that people with metabolic disease drink diet soda.

The counterweight is the older and larger NIH-AARP analysis. Lim and colleagues followed 285,079 men and 188,905 women aged 50 to 71 and recorded 1,888 hematopoietic cancers and 315 malignant gliomas. Higher aspartame intake was not associated with either. That is roughly 474,000 people against 103,000, looking at the specific cancers the original rodent scare pointed at, and finding nothing.

So the epidemiology is mixed, the effect sizes where they exist are small, and the exposure measurement is imperfect in every study on both sides. That is a genuinely unsatisfying state of evidence, and pretending otherwise in either direction is the thing this site exists not to do.

The Insulin Question, Settled and Unsettled

The claim circulates in every gym: sweet taste triggers an insulin response even without sugar, the insulin drives fat storage or hunger, and therefore diet soda makes you fat. It is a tidy mechanism and it has been tested more thoroughly than almost any other claim in this article.

Greyling and colleagues performed the definitive pooling in 2020, screening the literature through January of that year and including 26 papers covering 34 postprandial glucose trials and 29 postprandial insulin trials. The pooled effect of low-energy sweeteners on postprandial glucose was -0.02 mmol/L, with a confidence interval of -0.09 to 0.05. The pooled effect on postprandial insulin was -2.39 pmol/L, confidence interval -11.83 to 7.05. Both straddle zero and neither is clinically meaningful. The results did not differ by the type of sweetener, the dose, whether it was consumed alone or with a meal, or baseline glucose and insulin levels. Among people with type 2 diabetes, the glucose response was actually slightly lower after sweetener exposure, by 0.3 mmol/L.

One honest caveat: that meta-analysis was funded by Unilever, and two of the four authors were Unilever employees at the time. Industry funding does not make findings false, and the analysis was preregistered on PROSPERO, but it belongs in your assessment. The reassuring news is that it does not stand alone.

Grotz and colleagues ran a double-blind randomized trial in which normoglycemic men took roughly 333 mg of encapsulated sucralose three times daily at mealtimes for twelve weeks, a dose far above habitual intake and delivered in capsules to remove taste from the equation entirely. Fasting glucose, insulin, C-peptide, and HbA1c showed no significant change against placebo, and oral glucose tolerance tests showed no meaningful shift in time-to-peak or return toward baseline.

Higgins and Mattes gave 154 adults with overweight or obesity between 1.25 and 1.75 liters of sweetened beverage every day for twelve weeks, randomized across sucrose, aspartame, saccharin, sucralose, and rebaudioside A. Glucose tolerance was not significantly affected by any of the sweetener treatments.

Against that, the study everyone cites on the other side. Pepino and colleagues gave seventeen people with obesity, average BMI 42, who did not habitually use non-nutritive sweeteners, sucralose or water ten minutes before a glucose load. The sucralose condition produced roughly a 20 percent higher total insulin response. That finding is real and has never been retracted. It is also a single acute crossover in seventeen sweetener-naive people with severe obesity, testing sucralose combined with a glucose load rather than by itself, and it was contested in the journal at the time.

Put the whole file together and the defensible position is this. Sweeteners consumed on their own do not produce a meaningful insulin response in the pooled data. There may be a modest effect when a sweetener precedes a large carbohydrate load in people unaccustomed to sweeteners and carrying substantial obesity. That is a narrow and specific claim, and it is a long way from the gym version.

The Microbiome Question, Which Is Genuinely Open

Here the evidence is thinner and the mechanism is more plausible, which is an uncomfortable combination and the reason this section is the longest honest part of the article.

The modern line starts with Suez and colleagues in Nature in 2014, which showed glucose intolerance in mice fed saccharin, aspartame, or sucralose, demonstrated the effect disappeared when the animals were given antibiotics, and reproduced it by transplanting feces from sweetener-fed mice into germ-free mice. The human portion of that paper was much smaller: an observational slice of 381 non-diabetic individuals plus a brief saccharin intervention in a handful of healthy volunteers, in which roughly half showed poorer glycemic responses. It generated enormous coverage on the strength of the mouse work.

The same group came back in Cell in 2022 with the trial that should be the reference point. They randomized 120 healthy adults to two weeks of saccharin, sucralose, aspartame, or stevia in sachets at doses below the acceptable daily intake, against controls receiving the sachet vehicle glucose or nothing at all. Every sweetener distinctly altered the stool microbiome, the oral microbiome, and the plasma metabolome. Saccharin and sucralose significantly impaired glycemic responses. And critically, stool from human responders transplanted into germ-free mice reproduced the donors' glycemic patterns, which establishes that the microbiome shift was causing the glucose effect rather than merely accompanying it.

That is a well-designed trial and it deserves to be taken seriously. Three things keep it from being a verdict.

It ran for two weeks. A two-week shift in bacterial composition tells you the system is responsive, not that the shift persists, progresses, or produces anything clinically relevant. Second, the outcome was a glucose curve, not a disease. Impaired glycemic response over fourteen days in healthy adults is a biomarker, and the history of nutrition science is littered with biomarkers that never cashed out into outcomes. Third, and most interesting, the effects were strongly individual. The paper is titled around personalization for a reason: there were responders and non-responders within every sweetener group, which means population-level advice is exactly the wrong shape for this finding.

It is also worth noting where stevia sits in this picture, because the plant-derived framing implies it bypasses the issue and it does not. Steviol glycosides are not hydrolyzed by digestive enzymes in the upper gut and are not absorbed there. They travel to the colon intact, where Bacteroides species strip the glucose units and release steviol, which is then absorbed, glucuronidated in the liver, and excreted in urine. Stevia's entire metabolic route runs directly through your gut bacteria. Whatever the microbiome question turns out to be, stevia is inside it, not outside it.

Do They Actually Help You Lose Fat?

The correct answer is that they remove calories, and removing calories is the only mechanism on offer. A sweetener is not a fat-loss agent. It is a substitution tool, and its value is exactly the size of the substitution.

Rogers and colleagues pooled nine sustained intervention comparisons running from four weeks to forty months and found that low-energy sweetener consumption versus sugar produced about 1.35 kg less body weight, with a confidence interval from -2.28 to -0.42. That is a real effect and a modest one, and it comes entirely from the sugar that was not consumed.

Higgins and Mattes provides the cleanest picture because it ran five arms head-to-head in the same trial. Across twelve weeks of 1.25 to 1.75 liters daily, the sucrose group gained 1.85 kg and the saccharin group gained 1.18 kg, and those two did not differ from each other. Aspartame and rebaudioside A produced no significant change. Sucralose trended negative and was significantly lower than every other sweetener at week 12, with a difference of at least 1.37 kg, and energy intake actually fell in the sucralose arm.

That saccharin result is the interesting one, and the authors' conclusion is the line to take away: low-calorie sweeteners should be categorized as distinct entities because their effects on body weight differ. Lumping saccharin, sucralose, aspartame, and stevia into one category called artificial sweeteners is a habit of headline writers, not a fact about the molecules.

The practical translation for a lifter is narrow and worth stating precisely. If you currently drink 400 calories a day of sugar-sweetened beverages, switching to diet versions removes 400 calories a day, and that is worth about 0.4 kg of fat per week at face value before any compensatory eating. If you currently drink water, switching to diet soda removes nothing and does nothing. The sweetener never had a metabolic effect to offer. It had a subtraction to offer, and subtraction only works if there was something there to subtract. The same logic applies to every part of a cut, which is why tracking what you actually eat remains the thing that decides outcomes.

Why the WHO Guideline Says Something Different

In 2023 the WHO published a guideline advising that non-sugar sweeteners not be used as a means of achieving weight control or reducing the risk of noncommunicable diseases. That headline circulated widely and is usually quoted without the two qualifiers that came attached to it.

The first qualifier is the word conditional. In WHO's own grading vocabulary, a conditional recommendation means the guideline development group lacks confidence that the desirable consequences outweigh the undesirable ones. It is an explicit signal of uncertainty, and WHO says so in the document.

The second qualifier is low certainty. The underlying systematic review graded the evidence as low certainty overall. Short-term randomized trials, mostly three months or under, showed reduced sugar intake, reduced energy intake, lower body weight, and lower BMI with sweetener use, all at low certainty. The longer trials, six to eighteen months, did not show a body weight effect and were described as difficult to interpret because of differences in how they were conducted and reported. The signals for type 2 diabetes, cardiovascular disease, and mortality came from long-term observational data, which is exactly the setting where the reverse-causation problem described earlier is at its worst.

So the guideline is not a contradiction of the trial evidence. It is a judgment call about what to recommend at a population level when the short-term randomized evidence points one way, the long-term observational evidence points another, and everything is graded low certainty. Reasonable people disagreed with it in print, including a group publishing in the European Journal of Clinical Nutrition arguing it needed reconsideration.

For an individual lifter, the guideline's actual practical content is unobjectionable: sweeteners are not a health intervention, and drinking diet soda is not a strategy for anything except not drinking regular soda. That was already true.

The Four Sweeteners a Lifter Actually Meets

Read the back of a protein powder, a pre-workout, a zero-calorie electrolyte mix, and a sugar-free syrup and you will meet the same short list over and over. Here is what distinguishes them.

Sucralose

Six hundred times sweeter than sugar, heat stable, and therefore the workhorse of the supplement industry. It is the sweetener most likely to show up in your protein powder and your pre-workout. Its ADI of 5 mg per kilogram is the tightest of the mainstream options in milligram terms, which is a function of how the toxicology shook out rather than a red flag, but it does mean a heavy supplement user accumulates a larger fraction of the ceiling here than with aspartame. It is also the sweetener with the strongest human data in both directions: Grotz found nothing over twelve weeks, Higgins and Mattes found the most favorable body weight result of any arm, and Suez found it among the two that shifted glycemic responses.

One more thing about sucralose worth knowing. The tabletop product is mostly not sucralose. A Splenda packet is roughly 99 percent bulking agent, usually dextrose and maltodextrin, which is why it lists about 4 calories and a gram of carbohydrate. This is irrelevant at two packets and not irrelevant at twenty.

Aspartame

Two hundred times sweeter than sugar, not heat stable, so it disappears from baked goods and dominates cold beverages. It is a dipeptide of two amino acids, phenylalanine and aspartic acid, joined with a methyl ester. That composition is the source of both its harmlessness and its single real contraindication: people with phenylketonuria cannot metabolize phenylalanine efficiently and must restrict it, which is why products carry the phenylalanine warning. Newborns are screened for PKU before leaving the hospital, so if you do not know you have it, you almost certainly do not.

Aspartame does contain calories, roughly 4 per gram like any protein fragment, but at 200 times the sweetness of sugar the amounts used are trivial enough that products round to zero.

Acesulfame potassium

Two hundred times sweeter, heat stable, and almost always blended with something else, usually sucralose or aspartame, because on its own it carries a bitter edge at higher concentrations. The blend is why your protein powder ingredient list frequently shows two sweeteners. Ace-K was the second-named compound in the NutriNet-Sante cancer signal, at a hazard ratio of 1.13, with all the same caveats that apply to that cohort.

Stevia (steviol glycosides)

Two hundred to four hundred times sweeter, plant-derived, and regulated differently from the other three in a way that matters more than most people realize. See the next section.

The Natural Label and What It Hides

Stevia is marketed on a contrast: the natural one, the plant one, the one that is not a chemical. Three facts complicate that framing, and none of them makes stevia bad.

First, the regulatory path is different and it is arguably less stringent, not more. Aspartame, sucralose, acesulfame potassium, saccharin, neotame, and advantame are all approved food additives, meaning each went through FDA premarket review and approval. Purified steviol glycosides did not. They arrived through the GRAS notification route, in which a company concludes a substance is generally recognized as safe and notifies the FDA, and the FDA responds that it has no questions. The agency's language is precise about this: it has "not questioned the notifiers' GRAS conclusions." That is a different sentence from "FDA has approved."

Second, the plant itself is not permitted. Whole-leaf stevia and crude stevia extracts are subject to an FDA import alert and are not allowed for use as sweeteners in the United States. Only the highly purified steviol glycosides, at a minimum of 95 percent purity, have cleared the GRAS process. The product in your packet is an industrial extraction and purification of a specific set of molecules from a leaf, which is a perfectly reasonable thing to be and is not what most people picture when they read natural.

Third, the margin is narrower than aspartame's. The JECFA ADI is 4 mg per kilogram expressed as steviol equivalents, which the FDA converts to 12 mg per kilogram of rebaudioside A, giving 27 packets a day for a 60 kg person against aspartame's 75. Stevia has a smaller cushion, not a larger one.

Monk fruit sits in a similar place with a friendlier number. Siraitia grosvenorii fruit extract also came through GRAS notification, is 100 to 250 times sweeter than sugar, and has no ADI specified at all, because the evidence indicated safety well above the amount needed to sweeten anything.

Using Sweeteners on a Cut Without Fooling Yourself

Here is where the practical value sits, and it is real. A cut is an exercise in staying in a deficit long enough for it to work, and the failure mode is almost never metabolic. It is quitting. Anything that makes a deficit more tolerable at no caloric cost is worth having.

Diet soda as a hunger tool has limits. Carbonation and cold produce a short-lived fullness that can carry you through the forty minutes between a craving and your next meal. It does not replace the actual levers, which are protein, fiber, and food volume. If you are relying on diet soda to survive an aggressive deficit, the deficit is too aggressive, and a planned diet break is a better answer than another can.

Sugar-free syrups and zero-calorie condiments are the most underrated item on this list. Zero-calorie hot sauce, mustard, sugar-free ketchup, and low-calorie syrups let you make dull cut food palatable without spending anything from the budget. Palatability drives adherence, and adherence decides cuts.

Protein powder is the point of contact most lifters forget. Nearly every flavored powder on the market is sweetened with sucralose, acesulfame potassium, or stevia. If you drink two shakes a day, you are a regular sweetener consumer whether or not you think of yourself as one. This is worth noting mainly so your mental accounting is accurate. Our protein powder buyer's guide covers what else to look for on those labels.

Do not build a compensation story. The one behavioral effect worth watching has nothing to do with physiology. It is the mental accounting where a zero-calorie choice buys permission for a caloric one: diet soda with the large fries. The sweetener did not cause the fries. The reasoning did. If you notice yourself doing this, the fix is tracking, not abstinence.

Watch the sweetness setpoint, carefully and without dogma. There is a common claim that heavy sweetener use raises your threshold for sweetness and makes real food taste bland. The controlled human evidence for that is weak. But some people report it and it is cheap to test: if everything unsweetened tastes like nothing to you, dial back for a few weeks and see. That is a self-experiment, not a finding, and it should be labeled as one.

The Gut Problem Nobody Blames Correctly

When someone tells you a zero-calorie product wrecks their stomach, the culprit is usually not the high-intensity sweetener. It is the sugar alcohol sitting next to it.

Sorbitol, xylitol, lactitol, mannitol, erythritol, and maltitol are a different class entirely. They are 25 to 100 percent as sweet as sugar, not hundreds of times sweeter, so they are used in gram quantities rather than milligram quantities. They are incompletely absorbed in the small intestine, which means they arrive in the colon in bulk, pull water osmotically, and get fermented. The result is bloating, gas, and for some people an urgent trip. This is why sugar-free candy, protein bars, and keto desserts have the reputation they do, and why a diet soda sweetened with aspartame does not.

Erythritol is the partial exception, since it is largely absorbed in the small intestine and excreted unchanged, making it the best tolerated of the group at moderate doses. It caught a headline of its own in 2023 when a paper in Nature Medicine linked it to cardiovascular events. The FDA reviewed that paper and concluded the observational studies it cited did not establish a causal link, while saying it would continue to monitor new information. Same pattern as everything else on this page: an association in observational data, a mechanism proposed, and a causal claim that outran the evidence.

The practical rule is to read the ingredient panel and attribute correctly. If a product gives you gut trouble, look for something ending in -ol before you blame the sucralose.

Who Actually Has a Reason to Avoid Them

Most of this article argues that the standard fears do not survive the dose arithmetic. That is not the same as saying nobody should avoid these compounds. Four groups have a real reason.

People with phenylketonuria. This is the only hard medical contraindication in the entire category, and it applies specifically to aspartame because of its phenylalanine content. It is also the reason the warning statement exists on labels.

People who get reproducible gastrointestinal symptoms. Usually sugar alcohols, as covered above, but occasionally a specific high-intensity sweetener. If you can reproduce it across three separate exposures, that is your body's data and it outranks any trial.

People whose intake is genuinely extreme. The person putting eight packets in every coffee, drinking a liter of diet soda, and taking three sucralose-sweetened supplements is nowhere near the aspartame ceiling but is far closer to the sucralose one than they imagine. Vary your sources or cut the total.

People for whom the uncertainty itself is the cost. If the microbiome literature bothers you, you will not enjoy any of this, and worrying about your diet soda every day has a real price. The reasonable response to a genuinely open question is not forced consumption. Drink water, use fruit, and spend your attention on the variables that are settled.

Common Mistakes

Treating an IARC group as a risk estimate

Group 2B means the evidence is limited, not that the risk is established or the exposure is relevant. The same classification system holds aloe vera extract and pickled vegetables. Read hazard classifications as statements about evidence quality, not danger levels.

Quoting the WHO guideline without the qualifiers

Conditional, low certainty, and explicitly not a safety statement. The guideline advises against using sweeteners as a weight-control strategy, which is different from advising against consuming them, and the document itself says the group lacked confidence in the balance of consequences.

Lumping all sweeteners together

Higgins and Mattes found saccharin behaving like sucrose on body weight while sucralose did the opposite in the same trial. Suez found saccharin and sucralose shifting glycemic responses while aspartame and stevia did not in the same design. The category is a shelf label, not a mechanism.

Assuming plant-derived means less processed or better vetted

Purified steviol glycosides came through GRAS notification rather than FDA premarket approval, the crude leaf extract is banned outright, and the daily ceiling in packets is roughly a third of aspartame's. None of that makes stevia bad. It makes the natural framing unearned.

Blaming the sweetener for the sugar alcohol

The gram-scale polyols cause the gut symptoms. The milligram-scale high-intensity sweeteners generally do not. Check the panel before you draw a conclusion.

Believing a sweetener can drive fat loss

It cannot. It can only remove calories you were otherwise going to drink. Swapping water for diet soda accomplishes precisely nothing, and no amount of sweetener fixes a diet that is not in a deficit.

Frequently Asked Questions

Does aspartame cause cancer?

The two WHO bodies that looked at it in July 2023 gave answers that sound contradictory and are not. IARC placed aspartame in Group 2B, possibly carcinogenic to humans, on the basis of limited evidence for liver cancer in humans, limited evidence in animals, and limited mechanistic evidence. On the same day, JECFA reviewed the same body of work and left the acceptable daily intake at 0 to 40 mg per kilogram of body weight. IARC answers whether a substance could ever cause cancer under some exposure. JECFA answers whether it does at the amounts people actually consume. The FDA went further and said it disagrees with the IARC classification, citing what it called significant shortcomings in the studies IARC relied on. For a lifter drinking one or two diet sodas a day, the practical answer is that no regulator anywhere changed a single intake number in response to this.

How much diet soda is actually too much?

Run the arithmetic on your own body weight. The FDA acceptable daily intake for aspartame is 50 mg per kilogram per day and the JECFA figure is 40. A can of diet soda carries roughly 200 to 300 mg of aspartame. A 90 kg lifter therefore has a ceiling of 4,500 mg by the FDA number or 3,600 mg by the JECFA number, which works out to somewhere between roughly 12 and 22 cans per day depending on which limit and which formulation you use. WHO made the same calculation for a 70 kg adult and landed on more than 9 to 14 cans daily. Those ceilings already include a safety factor, typically a hundredfold margin below the highest dose that produced no observed effect in animal testing. So ordinary diet soda drinking is nowhere near the limit, and the better question is not whether two cans is dangerous but whether the habit is displacing water and food you would otherwise be better off with.

Do artificial sweeteners spike insulin?

The pooled human evidence says no. Greyling and colleagues meta-analyzed 34 postprandial glucose trials and 29 postprandial insulin trials and found a pooled glucose difference of -0.02 mmol/L and a pooled insulin difference of -2.39 pmol/L against control, neither meaningfully different from zero, with no difference by sweetener type or dose. Grotz and colleagues gave healthy men about 1,000 mg of encapsulated sucralose a day for twelve weeks and saw no change in fasting glucose, insulin, C-peptide, or HbA1c. The trial quoted on the other side is Pepino 2013, which found a roughly 20 percent higher insulin peak after a glucose load in seventeen people with obesity who did not normally use sweeteners. That result is real but narrow, and it involves sucralose taken alongside a glucose load rather than on its own. Sweetener in your coffee is not putting you in a fed state.

Do artificial sweeteners break a fast?

If your definition of the fast is calories and insulin, a zero-calorie sweetener does not break it, and the pooled insulin data is the reason. A packet of sucralose or a diet soda delivers no meaningful energy and no measurable insulin response. If your definition is a digestive or autophagy-related rest, nobody has the human data to answer that, and anyone who tells you otherwise is extrapolating from cell and rodent work. Two practical footnotes. Tabletop packets contain a bulking agent, usually dextrose or maltodextrin, which is why a packet of Splenda lists about 4 calories and 1 gram of carbohydrate even though the sucralose itself contributes almost nothing. Thirty packets is no longer zero. And flavored branched-chain amino acid drinks are not zero-calorie, because the amino acids themselves carry energy.

Is stevia safer than aspartame or sucralose?

Safer is the wrong frame, because all of them clear the same regulatory bar and none has a demonstrated harm at normal intakes. What is true is that stevia has a different regulatory path and a much narrower margin in packet terms. Purified steviol glycosides are not FDA-approved food additives at all. They sit under GRAS notices the FDA has declined to question, while whole-leaf and crude stevia extracts remain on an import alert and are not permitted as sweeteners. The JECFA acceptable daily intake is 4 mg per kilogram expressed as steviol equivalents, which the FDA translates to 12 mg per kilogram of rebaudioside A. By the FDA's own packet arithmetic that is 27 packets a day for a 60 kg person against 75 for aspartame. Steviol glycosides also pass the small intestine intact and are hydrolyzed to steviol by Bacteroides in the colon, so the plant-derived label does not mean your gut never meets the molecule.

Do artificial sweeteners wreck your gut microbiome?

They change it. Whether that change matters for your health is the part nobody has demonstrated. Suez and colleagues randomized 120 healthy adults to two weeks of saccharin, sucralose, aspartame, or stevia at doses below the acceptable daily intake and found that each one distinctly altered the stool and oral microbiome and the plasma metabolome, while only saccharin and sucralose measurably impaired glycemic responses. Transplanting stool from human responders into germ-free mice reproduced the glycemic pattern, which is genuine causal evidence for the mechanism. What that trial did not show is any clinical outcome. Two weeks of altered bacterial composition and a shifted glucose curve is not a disease, and the effect was strongly individual, with responders and non-responders inside every group. Treat this as an open question rather than a settled verdict in either direction.

Will diet soda help me lose fat?

Only by subtraction, and only if it actually replaces sugar. Rogers and colleagues pooled nine sustained comparisons and found low-energy sweeteners produced about 1.35 kg less body weight than sugar over periods running from four weeks to forty months. Higgins and Mattes ran the cleanest head-to-head, giving 154 adults with overweight or obesity 1.25 to 1.75 liters of sweetened beverage daily for twelve weeks, and found the sucrose arm gained 1.85 kg while the aspartame, rebaudioside A, and sucralose arms did not change significantly. Saccharin was the odd one out, gaining 1.18 kg. The WHO guideline reaches a different conclusion because it asks a different question, leaning on long-term observational data where reverse causation is severe. None of this makes a sweetener a fat-loss agent. It removes calories you were drinking. If you were already drinking water, switching to diet soda does nothing at all.

The Bottom Line

The cancer question has been answered as well as this kind of question gets answered. IARC's Group 2B says the evidence that aspartame could cause cancer under some exposure is limited, JECFA says the intake required to matter is far beyond what anyone consumes, the FDA says it disagrees with IARC's reading of the studies, and no regulator moved a number. The larger of the two major cohorts, at roughly 474,000 people, found nothing for the cancers the original scare pointed at. The smaller one, at 103,000, found hazard ratios around 1.13 with intervals brushing 1.00 in a setting where reverse causation is the default explanation.

The insulin question is closed in the direction of no effect. Thirty-four glucose trials and 29 insulin trials pooled to essentially zero, a twelve-week high-dose encapsulated sucralose trial found nothing, and a twelve-week five-arm beverage trial found no effect on glucose tolerance from any sweetener. The single contrary acute finding involves a sweetener-naive population with severe obesity consuming sucralose immediately before a glucose load, which is a real result about a narrow situation.

The microbiome question is open and should be described that way. The best trial found real, causally demonstrated, highly individual changes over fourteen days with no clinical endpoint. That is neither an all-clear nor an indictment, and stevia is inside that question rather than exempt from it, because gut bacteria are its metabolic pathway.

The weight question resolves to arithmetic. Sweeteners subtract calories you were drinking and do nothing else. About 1.35 kg against sugar in pooled sustained trials, and different sweeteners behaving differently in the one trial that ran them head-to-head. If you drink water already, switching changes nothing.

So use them if they make a cut easier, which for most people they do. Vary your sources rather than routing everything through sucralose. Read the panel for sugar alcohols before blaming a sweetener for your stomach. Skip aspartame if you have PKU. And if the open microbiome question bothers you enough to think about daily, drink water and stop reading about it, because the anxiety costs more than the compound does.

What none of this earns is a moral frame. The lifter drinking two diet sodas a day is not cheating, and the lifter who drinks none is not purer. Both are doing something that rounds to zero next to the variables that actually decide the outcome, which are the total calories, the protein, the training, and whether they keep doing it. Those are boring, which is why they never get a press release.

References

1. World Health Organization (2023). Aspartame hazard and risk assessment results released. Joint IARC and JECFA statement, 14 July 2023. Available at who.int.
2. International Agency for Research on Cancer (2023). Aspartame, Methyleugenol, and Isoeugenol. IARC Monographs on the Identification of Carcinogenic Hazards to Humans, Volume 134.
3. US Food and Drug Administration (2025). Aspartame and Other Sweeteners in Food, including "FDA Response to External Safety Reviews of Aspartame" and the Safe Levels of Sweeteners chart. Content current as of 27 February 2025.
4. US Food and Drug Administration (2017). High-Intensity Sweeteners. Questions and answers on regulatory status, GRAS notification, and acceptable daily intake.
5. Debras, C., Chazelas, E., Srour, B., Druesne-Pecollo, N., Esseddik, Y., Szabo de Edelenyi, F., et al. (2022). Artificial sweeteners and cancer risk: Results from the NutriNet-Sante population-based cohort study. PLOS Medicine, 19(3), e1003950. doi:10.1371/journal.pmed.1003950
6. Lim, U., Subar, A.F., Mouw, T., Hartge, P., Morton, L.M., Stolzenberg-Solomon, R., et al. (2006). Consumption of aspartame-containing beverages and incidence of hematopoietic and brain malignancies. Cancer Epidemiology, Biomarkers & Prevention, 15(9), 1654-1659. doi:10.1158/1055-9965.EPI-06-0203
7. Greyling, A., Appleton, K.M., Raben, A., & Mela, D.J. (2020). Acute glycemic and insulinemic effects of low-energy sweeteners: a systematic review and meta-analysis of randomized controlled trials. The American Journal of Clinical Nutrition, 112(4), 1002-1014. PROSPERO CRD42018099608. Funded by Unilever.
8. Grotz, V.L., Pi-Sunyer, X., Porte, D., Roberts, A., & Richard Trout, J. (2017). A 12-week randomized clinical trial investigating the potential for sucralose to affect glucose homeostasis. Regulatory Toxicology and Pharmacology, 88, 22-33. doi:10.1016/j.yrtph.2017.05.011
9. Pepino, M.Y., Tiemann, C.D., Patterson, B.W., Wice, B.M., & Klein, S. (2013). Sucralose affects glycemic and hormonal responses to an oral glucose load. Diabetes Care, 36(9), 2530-2535. doi:10.2337/dc12-2221
10. Higgins, K.A., & Mattes, R.D. (2019). A randomized controlled trial contrasting the effects of 4 low-calorie sweeteners and sucrose on body weight in adults with overweight or obesity. The American Journal of Clinical Nutrition, 109(5), 1288-1301. doi:10.1093/ajcn/nqy381. Registered NCT02928653.
11. Rogers, P.J., Hogenkamp, P.S., de Graaf, C., Higgs, S., Lluch, A., Ness, A.R., et al. (2016). Does low-energy sweetener consumption affect energy intake and body weight? A systematic review, including meta-analyses, of the evidence from human and animal studies. International Journal of Obesity, 40(3), 381-394. doi:10.1038/ijo.2015.177
12. Suez, J., Korem, T., Zeevi, D., Zilberman-Schapira, G., Thaiss, C.A., Maza, O., et al. (2014). Artificial sweeteners induce glucose intolerance by altering the gut microbiota. Nature, 514(7521), 181-186. doi:10.1038/nature13793
13. Suez, J., Cohen, Y., Valdes-Mas, R., Mor, U., Dori-Bachash, M., Federici, S., et al. (2022). Personalized microbiome-driven effects of non-nutritive sweeteners on human glucose tolerance. Cell, 185(18), 3307-3328. doi:10.1016/j.cell.2022.07.016
14. World Health Organization (2023). Use of non-sugar sweeteners: WHO guideline. Geneva: World Health Organization. Conditional recommendation, low certainty evidence.
15. Mela, D.J., McLaughlin, J., & Rogers, P.J. (2023). WHO guideline on the use of non-sugar sweeteners: a need for reconsideration. European Journal of Clinical Nutrition, 77, 1023-1025. doi:10.1038/s41430-023-01314-7
16. Wheeler, A., Boileau, A.C., Winkler, P.C., Compton, J.C., Prakash, I., Jiang, X., & Mandarino, D.A. (2008). Pharmacokinetics of rebaudioside A and stevioside after single oral doses in healthy men. Food and Chemical Toxicology, 46(Suppl 7), S54-S60. doi:10.1016/j.fct.2008.04.041
17. Purkayastha, S., & Kwok, D. (2020). Metabolic fate in adult and pediatric population of steviol glycosides produced from stevia leaf extract by different production technologies. Regulatory Toxicology and Pharmacology, 116, 104727. doi:10.1016/j.yrtph.2020.104727
18. US Food and Drug Administration (2023). FDA Memo: An evaluation of the article "The artificial sweetener erythritol and cardiovascular event risk" by Witkowski et al., Nat Med. 2023 Mar;29(3):710-718.