An evidence-graded field guide

Sweet Nothings

Artificial sweeteners, sugar alcohols, rare sugars, and the synthetic dyes that color them: what the trials actually show, what the cohort studies can't tell you, and the tools to work out what any of it means for your own diet.

The pink/blue/yellow/green packet colors were never regulated, just claimed one by one from 1957 onward, and are now legally protected trade dress in the US. Everything on this page is cited; the numbers in every tool trace to a source in the references. Nothing here is medical advice.

If you read one screen

Sweet Nothings, in one screen

sept 2026
  • Diet soda beats sugar soda. In every randomized trial. Water beats both, barely.
  • The scary studies watch who chooses diet. People already gaining weight. Trials that assign sweeteners don't find the harm.
  • Aspartame: "possibly carcinogenic" and "safe at 9–14 cans a day." Same day, same WHO. Hazard flagged, risk not found.
  • Sugar alcohols' real side effect is the bathroom. Sorbitol, maltitol from ~15–40 g; erythritol ~50 g. Tolerance builds in two weeks.
  • Erythritol and clots: unresolved. A 30 g platelet signal, no outcome trial, and your liver makes it too. Skip it if high-risk.
  • Best taste: allulose. Best in the oven: allulose. Erythritol goes gritty; aspartame breaks down with heat.
  • Dyes are banned for law and behavior, not cancer. Red 3 fell to a rat-only rule; the EU labels six for small effects in some kids.
Water
>
Diet soda
>
Sugar soda

sweetnothings · Reid Robison MD · every claim cited below

The cast

Three families, and the family matters more than the label. "Artificial" tells you almost nothing; "how much of it reaches your colon" tells you a lot.

High-intensity sweeteners

Hundreds to thousands of times sweeter than sugar, so used in milligrams. Mostly not absorbed (sucralose, ace-K, saccharin) or broken into ordinary dietary components (aspartame → aspartic acid, phenylalanine, methanol, in amounts smaller than a glass of tomato juice provides).1

Sweetener× sugarADI (mg/kg/day, JECFA)FoundNotes
Saccharin (the pink packet: Sweet'N Low)300–40051879Oldest. Bitter/metallic at high dose. Bladder-cancer scare (rat-specific mechanism) formally retracted; delisted from US carcinogen list in 2000.
Cyclamate30–5011 (EU 7)1937Banned in the US since 1970 on a bladder study never replicated; legal in the EU, Canada, and ~100 countries, often blended 10:1 with saccharin.
Aspartame (the blue packet: Equal)~20040 (FDA 50)1965Cleanest sugar-like taste of the classics. Loses sweetness with heat and long shelf time. Must carry a phenylketonuria warning.
Acesulfame K~200151967Heat stable, slight bitterness, almost always blended (Coke Zero = aspartame + ace-K). Excreted unchanged.
Sucralose (the yellow packet: Splenda)~60015 (FDA 5)1976Chlorinated sucrose; ~85% passes through unabsorbed. Heat stable. The sweetener most implicated in the human microbiome/glycemia trial.6
Neotame7,000–13,00022002Aspartame derivative; so potent that phenylalanine exposure is negligible and no PKU label is required.
Advantame~20,00052014Sweetest approved compound on Earth. Used in micrograms.
Steviol glycosides (the green packet: Truvia)200–3504 (as steviol)1931Reb A has a licorice/bitter tail; newer Reb M and Reb D are much cleaner. "Natural" origin, industrial purification.
Monk fruit (mogroside V)150–300GRAS, none set2010 (US)Fruity, mild aftertaste. Nearly always sold blended with erythritol for bulk. Not yet approved in the EU.

Sugar alcohols (polyols)

Bulk sweeteners used gram-for-gram like sugar. They share a chemistry (a sugar with its carbonyl reduced to a hydroxyl) and a side effect (what isn't absorbed pulls water into the colon and feeds bacteria). Absorption is the whole story: erythritol is ~90% absorbed in the small intestine and excreted in urine; sorbitol and maltitol are mostly not.11

Polyol× sugarkcal/gAbsorbedGI toleranceNotes
Erythritol0.6–0.70.2~90%Best of the groupCooling effect, recrystallizes. Occurs naturally in melon, pear, wine, and is made by the body (endogenous). Center of the 2023–24 clot debate.
Xylitol1.02.4~50%ModerateOnly polyol as sweet as sugar. Anti-caries evidence for gum. Lethal to dogs at small doses (insulin surge).
Maltitol0.92.1~40%PoorThe "sugar-free" candy default; the one behind most Amazon gummy-bear reviews. Raises blood sugar more than the others (GI ≈ 35).
Sorbitol0.62.6~25%PoorIn sugar-free gum, diet foods, and prunes (hence prunes). Laxative threshold is the lowest of the group.12
Mannitol0.51.6~25%PoorMostly pharma/dusting powder. US labels must warn of a laxative effect above 20 g/day.
Isomalt0.52.0~10%Poor–moderateThe pastry chef's polyol: pulls, blows, and casts like sugar without browning or going sticky.
Lactitol0.42.0~2%PoorMade from lactose; used in chocolate and, at 10–20 g doses, as an actual laxative.

Rare sugars

Allulose (D-psicose) is a real sugar, an epimer of fructose, that the body absorbs but can't metabolize: ~0.4 kcal/g, 70% as sweet as sucrose, browns and caramelizes in the oven, and is excreted in urine. Approved in the US, Japan, Korea, Mexico; GI tolerance sits between erythritol and xylitol (laxative threshold ≈ 0.5 g/kg single dose). Not yet approved in the EU. Tagatose is the galactose analogue, 92% as sweet, 1.5 kcal/g, less common.13

How sweet is sweet?

Potency relative to sucrose (= 1). The bars are on a log scale; a linear one would make everything but advantame invisible.

Potency is measured by sensory panels matching sweetness intensity at typical use levels and falls at high concentrations for most high-intensity sweeteners, which is why "600×" is a range, not a constant.

The science, claim by claim

Every big scare and every big reassurance comes from one of two study types that disagree with each other. Once you know why, the whole field gets less confusing.

The one idea to take away. Randomized trials, where people are assigned a sweetener, almost always show neutral-to-favorable effects on weight and glucose compared with sugar. Prospective cohorts, where researchers watch who chooses diet products, often show the opposite. The people who choose diet soda are disproportionately people already gaining weight, already prediabetic, already told to cut sugar. That is reverse causation, and the WHO itself named it as the reason its 2023 sweetener guideline was only a "conditional" recommendation built on "low to very low certainty" evidence.2,3 When cohorts are re-analyzed to model the intended substitution (sweetener instead of sugar) and adjust for baseline adiposity, the harm signals shrink or flip.4

"Aspartame causes cancer"hazard flagged, risk not shownThe 2023 IARC "possibly carcinogenic" headline, decoded.

In July 2023 two WHO bodies reported on the same day. IARC, which identifies hazards (could it, under any conditions), put aspartame in Group 2B on "limited" evidence for liver cancer from three cohorts, "limited" animal evidence (the Ramazzini Institute rat studies, whose methods both bodies criticized), and "limited" mechanistic evidence. JECFA, which assesses risk (does it, at real doses), reviewed the same data and reaffirmed the 40 mg/kg/day ADI, calling the human cancer evidence "not convincing."1

Group 2B is the category that also contains aloe vera extract, pickled vegetables, and working as a dry cleaner. It says nothing about dose. A 70 kg adult would need 9–14 cans of diet soda every day to reach the ADI, which itself carries a 100-fold safety factor from the no-effect level.1

The NutriNet-Santé cohort that drove the headlines found small associations at intakes 20–40× below the ADI, with no dose–response, which is the pattern you'd expect from confounding rather than carcinogenesis.5

Evidence: RCTs impossible for cancer; large cohorts mixed; regulatory consensus unchanged (FDA, EFSA, JECFA, Health Canada).

"Diet sweeteners make you fat"not when they replace sugarThe cohort/RCT split at its starkest.

The 2016 Rogers meta-analysis of sustained RCTs (4 weeks to 40 months) found sweeteners vs sugar reduced body weight by about 1.35 kg, and, in the handful of trials that compared them with water, sweeteners did as well or slightly better.7 A 2024 Obesity Reviews meta-analysis restricted to people who already drank sugary drinks found switching to non-caloric drinks lowered BMI by 0.31 kg/m² long-term.8 A 52-week Liverpool RCT of 493 adults in a weight-loss program found the diet-beverage group kept off 7.5 kg vs 6.1 kg for water, a statistically but not clinically meaningful edge (industry-funded; open-label).9

The 2025 Sievenpiper umbrella review then showed that when cohorts are analyzed the same way trials are (substitution or change analysis), the association with weight gain disappears and often reverses.4 The 2026 Tufts review still argues the direct physiological effects (independent of calorie replacement) deserve more study, and that sweetened beverages should be a bridge, not a destination.10

Evidence: Moderate certainty from RCTs that sweeteners reduce energy intake and weight vs sugar; low certainty for any harm.

"They spike insulin / confuse your body"acute effects are near zeroCephalic-phase insulin, GLP-1, and the "sweet taste without calories" theory.

A 2023 network meta-analysis of acute trials found non-nutritive sweetened drinks had no meaningful effect on postprandial glucose, insulin, GLP-1, or GIP compared with water, and blunted the responses compared with sugary drinks.14 The idea that sweet taste alone triggers a meaningful insulin release in humans has not held up outside of a few small saccharin studies from the 1980s.

The one real signal: a 2025 USC fMRI study found sucralose (vs sucrose or water) increased hypothalamic blood flow and reported hunger over ~2 hours, especially in people with obesity.27 It's an interesting brain-response finding, not a weight outcome, and the trials that measure actual intake keep finding people don't eat more afterward.7

Evidence: High certainty for no acute glycemic effect; mechanistic brain findings preliminary.

"They wreck your gut microbiome"real for some sweeteners, in some peopleThe best human trial in the field, and what it did and didn't show.

Suez et al. (Cell, 2022) randomized 120 sweetener-naïve adults to two weeks of saccharin, sucralose, aspartame, stevia, glucose, or nothing, at doses below the ADI. All four sweeteners shifted the stool and oral microbiome; only saccharin and sucralose worsened glucose tolerance, and transplanting the responders' stool into germ-free mice transferred the impairment.6 Response was highly individual: some people didn't budge.

Caveats that matter: two weeks, healthy volunteers, glucose-tolerance curves rather than disease, and other trials (e.g. high-dose saccharin for two weeks, aspartame/sucralose crossover) found no microbiome or glycemic change. Reviews of the clinical literature call the evidence inconsistent and mostly short-term.15 Sugar alcohols, by contrast, are fermented by gut bacteria, which is both why they cause gas and why some (xylitol, erythritol) look mildly prebiotic in small studies.

Evidence: Moderate for microbiome shifts with sucralose/saccharin; low for downstream health consequences.

"Erythritol and xylitol cause blood clots and heart attacks"a real signal, an unresolved fightThe Cleveland Clinic series (2023–24) and its critics.

Hazen's group found that patients with high fasting plasma erythritol had roughly twice the 3-year rate of major cardiac events; that erythritol added to blood in vitro increased platelet aggregation; and that 30 g of erythritol (a pint of keto ice cream) or 30 g of xylitol in healthy volunteers raised plasma levels ~1,000-fold and made platelets more reactive for hours, while glucose did not.16,17,18

The critics' case: the cohort patients were not asked what they ate. Erythritol is made inside the body from glucose via the pentose phosphate pathway, and endogenous production rises with obesity, insulin resistance, and diabetes, so high fasting erythritol may be a marker of metabolic disease rather than a cause. Platelet-aggregation assays are a surrogate; no trial has shown clots or events. A 2025 Cardiovascular Research review noted a genetic condition (pentosuria) that keeps a related polyol chronically elevated causes no harm, and that a Mendelian randomization study did not support causation.19 Hazen's group rejected that critique in print.20

Reasonable reading in 2026: unresolved, mechanistically plausible, no outcome trial. If you have established cardiovascular disease or clotting risk, there's no cost to choosing other sweeteners. For everyone else, a packet of stevia/erythritol blend contains ~1–2 g, not 30.

Evidence: Low certainty; large-dose acute physiology plus a confounded cohort. No RCT with clinical endpoints.

"Sweeteners raise diabetes and heart-disease risk"cohorts say yes, trials say noThe core of the WHO 2023 guideline and the response to it.

Long-term cohorts (NutriNet-Santé, Nurses' Health, others) associate high sweetener intake with type 2 diabetes and cardiovascular events. RCTs consistently show sweetened products beat sugar on every cardiometabolic risk factor, and the highest-quality weight-loss and diabetes-remission programs (DiRECT, Look AHEAD, PREVIEW) run on sweetener-flavored meal replacements.21 The WHO's own guideline flagged reverse causation as a likely contributor to the cohort signals.2

The honest position is not "proven safe" but "the trials that could show harm haven't, and the studies showing harm can't rule out that sick people choose diet products."

Evidence: Low/very low certainty for harm (WHO grading); moderate for benefit vs sugar.

"Stevia and monk fruit are natural, so they're safer""natural" is a purification storyOrigin says nothing about biology.

Commercial stevia is 95%+ purified rebaudiosides extracted with solvents and resin columns; the leaf itself is not approved in the US or EU because the crude extract contains compounds with unclear safety. Monk fruit sweetener is a mogroside concentrate. Both have good safety data at approved doses, and neither has a clear health advantage over aspartame or sucralose. What they do have is less human evidence overall, because they're newer. Stevia was one of the four sweeteners in the Suez trial and did not impair glucose tolerance.6

Evidence: Safety data adequate for approval; comparative superiority unsupported.

"Sugar-free gum and candy give you diarrhea"true, dose-dependent, and predictableThe one side effect nobody disputes.

Unabsorbed polyols are osmotic: they hold water in the bowel, then colonic bacteria ferment them into gas and short-chain fatty acids. It's the same mechanism as lactose intolerance, and like lactose it is a physics problem, not a disease. Laxative thresholds from the Japanese dose-ranging studies: erythritol ≈ 0.66 g/kg (men) to 0.80 g/kg (women); sorbitol ≈ 0.17–0.24 g/kg; maltitol ≈ 0.8 g/kg with occasional use but far better tolerated with daily use.12,22 Tolerance adapts within 1–2 weeks. Use the estimator below.

Evidence: High certainty, decades of dose-ranging trials.

Myths, sorted

Tap to open. Each verdict links back to the evidence above.

MythAspartame turns into formaldehyde and poisons you.
Chemistry with the numbers removed. Aspartame's methanol is oxidized to formaldehyde and then formate, exactly as the methanol from fruit, juice, and wine is. A can of diet soda yields ~20 mg methanol; a glass of tomato juice ~ 60–90 mg. Formaldehyde is a normal one-carbon metabolite present in every cell and is cleared in seconds; it does not accumulate from dietary methanol.1
MythSucralose is "chlorinated" like bleach or pesticides.
Chlorine atoms in a molecule and chlorine gas are different things. Table salt is chlorinated sodium. Sucralose's three chlorines are what stop enzymes from breaking it down, which is why ~85% leaves unabsorbed. The real sucralose questions are the microbiome findings and the fact that it degrades above ~120 °C in some conditions, not "chlorine."6
MythSaccharin causes bladder cancer.
It did, in male rats, for a reason that does not exist in humans. High-dose sodium saccharin forms crystalline precipitates in rat urine that damage the bladder lining; rat urine chemistry (high pH, high protein, high calcium phosphate) makes this possible. Humans don't form the precipitate. The 1977 warning label was removed in 2000 and saccharin was delisted by the US National Toxicology Program.5
Myth"Diet" soda makes you crave sugar and eat more later.
Repeatedly tested, repeatedly not found. Meta-analysis of 129 preload comparisons: a sweetened drink before an ad-libitum meal led to ~94 fewer total calories than a sugary drink, and no difference from water.7 Sweet-taste "preference" also doesn't ratchet up in the trials that measured it.
MythErythritol is basically poison now.
An unresolved 30-gram signal was reported as a settled milligram fact. See the ledger entry: high plasma erythritol is at least partly a marker of metabolic disease (your liver makes it), the acute studies used 30 g boluses, and the effect is on platelet assays, not clinical events. Prudent avoidance if you're high-risk is reasonable; panic isn't.19
MythStevia is just a leaf.
The green packet is a solvent-purified glycoside concentrate, ~200× sweeter than the leaf. Whole-leaf stevia is not an approved food additive in the US or EU. Purification is not a bad thing here; it's just not what "natural" implies.
MythSweeteners are fine for dogs.
Xylitol is not. Dogs release insulin in response to xylitol; 0.1 g/kg causes hypoglycemia and ~0.5 g/kg can cause liver failure. A few pieces of gum or a spoon of "keto" peanut butter can kill a small dog. Other sweeteners are not known to be dangerous to dogs. Keep xylitol products where dogs can't reach them.
MythThe WHO says don't use sweeteners.
The WHO said don't count on them for long-term weight control, in a conditional (weak) recommendation, based on low-certainty evidence, and explicitly excluded people with diabetes. It did not say they are unsafe; safety is a separate JECFA process that has cleared every one of them. Three years of substitution-analysis research have since weakened the cohort evidence the guideline leaned on.2,3,4
MythSugar is natural, so it's the healthier choice.
Free sugars are the one dietary component here with high-certainty evidence of harm at common intakes: weight gain, dental caries, and, via sugary drinks, type 2 diabetes. Every regulatory body that hedges on sweeteners recommends cutting sugar without hedging.2

How much is too much? An ADI calculator

The acceptable daily intake is the amount you could consume every day for life with no expected effect, set 100× below the highest dose with no effect in animals. Here's what it looks like in cans and packets.

Enter your weight and pick a sweetener.

Product contents are typical values from manufacturer disclosures and published assays (Diet Coke ≈ 180 mg aspartame per 355 mL; Coke Zero ≈ 87 mg aspartame + 46 mg ace-K; Splenda packet ≈ 12 mg sucralose; Sweet'N Low packet ≈ 36 mg saccharin; Equal packet ≈ 37 mg aspartame; Truvia packet ≈ 8 mg Reb A). JECFA ADIs used; FDA values differ for aspartame (50), sucralose (5), and saccharin (15).

The sugar-alcohol gut estimator

Will that keto brownie, that bag of sugar-free gummies, or that "no sugar" protein bar send you to the bathroom? A dose-per-kilogram model built on the published laxative-threshold trials.

Pick a sugar alcohol, enter the grams, and estimate.
How the model works (and why it's an estimate)

Each polyol gets a single-dose laxative threshold in g/kg from dose-ranging trials: erythritol 0.66 (M) / 0.80 (F)12, sorbitol 0.17 / 0.2412, maltitol ~0.8 (occasional use)22, xylitol ~0.35 (≈ 30 g in an adult, with 50–60 g/day tolerated after adaptation)11, mannitol ~0.3 (the FDA's 20 g/day warning line), isomalt ~0.4, lactitol ~0.3, allulose ~0.5. Your dose is divided by (threshold × body weight) to give a ratio; below ~0.35 symptoms are unusual, 0.35–0.7 brings gas and bloating in a sizable minority, 0.7–1.0 loose stools become common, and above 1.0 you've crossed the median laxative dose. Modifiers: habitual use raises tolerance ~30% (adaptation is well documented for maltitol and xylitol); spreading the dose over hours raises it ~40%; an empty stomach lowers it ~20%; IBS lowers it ~40% (polyols are the "P" in FODMAP). Individual variation is large, and mixed polyols add up.

Which ones taste best?

Sensory panels agree more than internet arguments suggest. Sugar has a fast onset, a clean peak, and a clean finish. Every substitute misses on at least one of those.

SweetenerSugar-likenessAftertasteWhat you notice
Allulose★★★★★noneClosest to sugar: same onset, slight cooling, ~70% strength. The current gold standard for clean taste.
Sucralose★★★★☆slow, lingering sweetnessVery sugar-like at low doses; sweetness that hangs on too long at high doses ("Splenda tail").
Aspartame★★★★☆mild, lingeringClean and slightly slow. Loses potency in warm or old soda, which is why flat Diet Coke tastes like nothing.
Erythritol★★★★☆cooling, not bitterThe cooling is endothermic dissolution (the same reason it's in "cooling" mints). Blends well; carries other sweeteners.
Xylitol★★★★☆mild coolingSame strength as sugar, similar mouthfeel, gentler cooling than erythritol.
Monk fruit★★★☆☆fruity, faintly melonPleasant but distinct; better in fruit/tea contexts than in chocolate or coffee.
Stevia (Reb M)★★★★☆slightThe newer glycoside most products have quietly switched to. Far cleaner than Reb A.
Stevia (Reb A)★★☆☆☆licorice, bitterSlow onset, long bitter/anise finish, worse at higher doses. The reason people think they hate stevia.
Acesulfame K★★★☆☆bitter/metallic aloneFast onset, which is why it's blended with slower aspartame or sucralose to mimic sugar's curve.
Maltitol★★★☆☆cleanTastes great in chocolate, which is exactly the problem: nothing warns you before 40 g.
Saccharin★★☆☆☆bitter, metallic~25% of people carry a TAS2R variant that makes saccharin markedly more bitter. Tab tasted like Tab.
Cyclamate★★★☆☆mildly bitterMasks saccharin's bitterness in the classic 10:1 blend; Canada's Sweet'N Low is cyclamate, the US one is saccharin.

Ratings summarize descriptive sensory literature and panel data for each compound at typical use levels; blends (e.g. erythritol + monk fruit, aspartame + ace-K) usually beat any single ingredient because they cover each other's temporal gaps. Bitterness of saccharin and ace-K is genetically variable (TAS2R31/TAS2R43).

Which ones bake well?

Sugar does five jobs in a recipe: sweetness, bulk, moisture retention, browning, and tenderizing. High-intensity sweeteners do exactly one of them. That's the whole baking problem in a sentence.

Browns and caramelizes

Allulose is the only sugar-free option that undergoes Maillard browning and caramelizes; it does so faster and darker than sucrose, so drop the oven ~15 °C / 25 °F and pull early. Maltitol and tagatose brown a little. Erythritol, xylitol, and all the high-intensity sweeteners don't brown at all.

Survives the oven

Heat stable: sucralose, ace-K, saccharin, stevia, monk fruit, all polyols, allulose. Aspartame breaks down at baking temperatures and loses sweetness, which is why there's no aspartame in baked goods. Sucralose can degrade above ~120 °C in prolonged dry heat; fine for most home baking.

Provides bulk and structure

Only the gram-for-gram sweeteners: erythritol, xylitol, allulose, maltitol, isomalt. Replacing a cup of sugar with a teaspoon of stevia gives you a dense, pale, dry puck. Granulated "cup-for-cup" sucralose or stevia products bulk with maltodextrin or erythritol, which is why they work.

Stays soft

Erythritol recrystallizes as things cool: gritty cookies, crunchy frosting, "cold" mouthfeel. Fix with a 3:1 erythritol:allulose blend or powdered erythritol. Allulose is hygroscopic and keeps things moist and chewy, sometimes too soft (cookies won't crisp). Xylitol holds moisture well; yeast can't ferment it, so it's a poor choice for bread.

Sugar work and candy

Isomalt is what pastry chefs pull, blow, and cast with: it doesn't crystallize or absorb humidity and stays glass-clear. Allulose makes proper caramel sauce and toffee. Erythritol candy turns to sand.

Ice cream and frozen

Sugar lowers the freezing point; without it you get an ice block. Allulose depresses freezing point about like sugar and makes scoopable ice cream. Erythritol makes it rock-hard; commercial keto pints add allulose, vegetable glycerin, or both.

Substitution tool

Enter the sugar amount and choose a recipe type.

Conversions use relative sweetness (erythritol 0.7, allulose 0.7, xylitol 1.0, maltitol 0.9, granulated cup-for-cup blends 1.0) and cap polyol doses against the gut thresholds above. One US cup of granulated sugar ≈ 200 g.

Diet soda vs sugar soda: the verdict

Scored on what has actually been measured in people, head to head, with water as the control everyone forgets.

Outcome
Sugar soda
Diet soda
Water
Calories per 355 mL can
~140 kcal (39 g sugar)
0–5 kcal
0
Weight, RCTs (vs sugar)
reference
−1.3 kg over months7
similar to diet9
Blood glucose & insulin, acute
large spike
≈ water14
reference
Type 2 diabetes risk, cohorts
↑ consistently, dose-dependent
↑ in naïve analyses; ≈ null when substitution-modeled4
lowest
Dental erosion
acid + sugar: caries and erosion
acid only: erosion, no caries fuel
none
Liver fat (NAFLD)
fructose load ↑ liver fat in RCTs
no effect shown
none
Gut microbiome
sugar shifts it too
sucralose/saccharin shift it; aspartame less so6
reference
Cancer
via obesity, indirectly
IARC 2B hazard, JECFA no risk at real doses1
none
Hydration
yes
yes (caffeine's diuresis is trivial at soda doses)
yes
Bone (phosphoric acid in colas)
weak association, both colas
weak association, both colas
none

The debate, silenced

Diet soda beats sugar soda. Not "probably," not "for now." On every outcome that has been randomized in humans, replacing a sugary drink with a diet one is neutral or better, and the calorie gap alone is ~50,000 kcal a year for a one-can-a-day habit. The only outcome where diet soda might match sugar soda is the microbiome, and there the evidence points at specific sweeteners in some people, not at "diet."7,8,4,21

Water beats both, but the gap between water and diet soda is small enough that the largest 1-year RCT couldn't show water winning on weight.9 The practical hierarchy: water or unsweetened drinks first; diet soda as the tool for getting off sugar soda; sugar soda as an occasional treat, not a beverage.

The remaining honest uncertainty is whether decades of sweetener use has effects that neither a 1-year trial nor a confounded cohort can see. That uncertainty is real. It is also not a reason to keep drinking the one with 39 g of sugar in it.

Food dyes: what's banned where, and why

Synthetic food colors are the additive category where the regulatory map is changing fastest and where the "why" is the most misunderstood. Almost none of the bans are about cancer.

What they are

The US certified colors ("FD&C" numbers) are petroleum-derived azo or triarylmethane dyes: cheap, stable, brilliant, and not absorbed much. "Petroleum-derived" describes their feedstock, not their toxicity; aspirin and vanillin can also start as petrochemicals. Each batch is chemically certified by the FDA before sale, which is what the "certified" means. Europe uses E-numbers for the same molecules and permits several the US never approved (and vice versa).

The dyes, one by one

Key: allowed allowed with warning label restricted use not permitted / revoked. US status as of September 2026; details in the tool below.

Where is it banned? Check a dye

Pick a dye and a place.

Why: the three actual reasons

1. Behavior in children

The 2007 Southampton trial (n = 297, double-blind, general-population kids aged 3 and 8–9) found small increases in hyperactivity from two dye-plus-sodium-benzoate mixtures.23 EFSA judged the evidence "limited," the effect small, and inconsistent across ages and mixes, and could not blame any single dye. The EU still required a warning label on the six colors in 2008.24 California's OEHHA 2021 review concluded synthetic dyes "can" affect behavior in some children, with wide individual variation.28 That is the strongest human evidence against these dyes, and it is a small, real, subgroup effect.

2. The Delaney Clause

A 1960 US law: any color additive shown to cause cancer in humans or animals, at any dose, cannot be approved. Red No. 3 causes thyroid tumors in male rats via a rat-specific hormone mechanism the FDA itself says does not apply to humans; the agency banned it in cosmetics in 1990 and, in January 2025, in food, explicitly citing Delaney and not human risk.25 Legal bright line, not a health finding.

3. Politics and precaution

Most of the 2025–26 US phase-out is voluntary, driven by the HHS "MAHA" agenda, state school-meal laws, and retailers' private deadlines rather than new science.26 Europe's historic national bans (Norway, Sweden, Austria, Finland before 1990s harmonization) were largely precautionary and were lifted when EU rules took over. The dyes replaced them with (beet, spirulina, paprika, turmeric, carmine) are safe but fade, cost more, and one of them is ground insects.

The current US picture, plainly

What should a parent actually do? If a child seems reactive to brightly colored foods, a two-week trial without synthetic dyes is cheap, harmless, and about as much evidence as the literature can offer for any individual kid. Synthetic colors must be declared by name on US labels, so you don't need to wait for 2027. For everyone else, the health case against dyes is far weaker than the health case against the sugar they're usually coloring.

A short, strange history

Nearly every artificial sweetener was discovered by a chemist licking his fingers. Nearly every dye scandal started with candy.

1856Mauve.

Eighteen-year-old William Perkin tries to synthesize quinine from coal tar, fails, and gets a purple stain. Synthetic dye chemistry, and within a decade the first coal-tar food colors, begin here.

1879Saccharin.

Constantin Fahlberg, working on coal-tar derivatives in Ira Remsen's lab at Johns Hopkins, notices his dinner roll is sweet: he hadn't washed his hands. He patents it without Remsen, who never forgives him.

1906The Pure Food and Drug Act.

Candy was being colored with lead chromate, mercury sulfide, and copper arsenite. Bernhard Hesse's review whittles ~80 coal-tar dyes down to seven approved colors. Roosevelt, a saccharin user with diabetes-adjacent doctors' orders, tells the chief chemist that anyone who says saccharin is harmful "is an idiot."

1937Cyclamate.

Illinois grad student Michael Sveda sets his cigarette on the bench, picks it back up, and it tastes sweet. Sold as Sucaryl, then as the sugar in Tab and Fresca.

1950Halloween.

Children across the US fall ill after eating orange candy colored with Orange No. 1 at far higher concentrations than ever tested. The FDA delists Orange 1, Orange 2, and Red 32, and the 1960 Color Additive Amendments follow, with the Delaney Clause riding along.

1957The pink packet.

Ben Eisenstadt of Cumberland Packing in Brooklyn, who had invented the sugar packet but failed to patent it, puts saccharin in pink paper with a treble clef and calls it Sweet'N Low. Diners get their first sweetener packet.

1965Aspartame.

G.D. Searle chemist James Schlatter, working on an ulcer drug, licks his finger to turn a page. Approved in 1974, withdrawn over the Searle data controversy, re-approved in 1981; NutraSweet's blue packet (Equal) arrives in 1982.

1967Acesulfame K.

Karl Clauss at Hoechst licks a finger. The pattern is now unmistakable.

1970Cyclamate banned in the US.

One rat study of a cyclamate/saccharin blend shows bladder tumors; never replicated. Still banned in the US, legal almost everywhere else. Saccharin gets a cancer warning label in 1977 (removed 2000).

1976Sucralose.

At Queen Elizabeth College, London, Shashikant Phadnis is told by his supervisor to "test" a chlorinated sugar. He hears "taste." Splenda's yellow packet reaches the US in 1999.

1976Red No. 2 and the missing red M&M.

A Soviet study and an FDA rat study raise cancer concerns about amaranth; the FDA bans it. Mars had never used Red 2 in M&Ms but pulls red ones anyway to avoid confusion. They return in 1987, colored with Red 40.

1990Red No. 3 loses cosmetics.

Male-rat thyroid tumors trigger Delaney for lipstick and external drugs. The FDA says it will follow in food; it takes 35 years.

2008Stevia goes mainstream.

Purified Reb A gets US GRAS status; Truvia's green packet launches. Erythritol, commercialized in Japan since 1990, rides along as its bulking partner.

2008The EU warning label.

After Southampton, six colors must carry "may have an adverse effect on activity and attention in children." UK Smarties, Skittles, and Fanta are quietly reformulated with natural colors; the US versions are not.

2015Allulose.

Tate & Lyle's Dolcia Prima brings the "rare sugar" to market; in 2019 the FDA rules it doesn't count as added sugar on labels.

2023The WHO year.

May: WHO advises against relying on sweeteners for weight control (conditionally). July: IARC calls aspartame "possibly carcinogenic" while JECFA, in the same press release, reaffirms the ADI. February: Hazen links erythritol to cardiac events.

2025Red 3 falls; the phase-out begins.

January: FDA revokes Red 3 under Delaney. April: HHS announces a voluntary end to the six remaining petroleum dyes by end of 2026. March: West Virginia passes the first broad state ban; a federal judge blocks its retail half in December.

2026Reformulation season.

Virginia's and Utah's school bans take effect; Texas warning labels begin; the federal phase-out deadline arrives with no rule behind it. Red 3's hard deadline is January 15, 2027.

Fun facts you didn't ask for: carmine (E120, "natural red 4") takes roughly 70,000 cochineal insects per pound; Starbucks dropped it from strawberry drinks in 2012 after a petition. Blue 1 is used in surgical dye and to trace lymph nodes. Erythritol was first isolated from algae in 1852. The cooling sensation of erythritol on the tongue is the heat of solution, measurable with a thermometer: about −43 kJ/kg, roughly three times xylitol's.

References

  1. IARC & JECFA. Aspartame hazard and risk assessment results released; joint summary of findings. WHO, 14 July 2023. who.int
  2. WHO. Use of non-sugar sweeteners: WHO guideline. Geneva, 2023. who.int
  3. Khan TA, Lee JJ, Ayoub-Charette S, et al. WHO guideline on the use of non-sugar sweeteners: a need for reconsideration. Eur J Clin Nutr 2023;77:1009–13. doi
  4. Ayoub-Charette S, Kavanagh M, Khan TA, Sievenpiper JL. Reconciling conflicting evidence on low- and no-calorie sweeteners and cardiometabolic outcomes: an umbrella review using naïve and bias-adjusted methods. Appl Physiol Nutr Metab 2025;50:1–26. doi
  5. National Cancer Institute. Artificial sweeteners and cancer (fact sheet). cancer.gov
  6. Suez J, Cohen Y, Valdés-Mas R, et al. Personalized microbiome-driven effects of non-nutritive sweeteners on human glucose tolerance. Cell 2022;185:3307–28. doi
  7. Rogers PJ, Hogenkamp PS, de Graaf C, et al. 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. Int J Obes 2016;40:381–94. PMC
  8. Tobiassen PAS, Køster-Rasmussen R. Substitution of sugar-sweetened beverages with non-caloric alternatives and weight change: a systematic review of randomized trials and meta-analysis. Obes Rev 2024;25:e13652. doi
  9. Harrold JA, Hill S, Radu C, et al. Non-nutritive sweetened beverages versus water after a 52-week weight management programme: a randomised controlled trial. Int J Obes 2024;48:83–91. doi
  10. Wang M, Wu OY, Wallen OG, Mozaffarian D. Artificial and other non-nutritive sweeteners, the microbiome, and cardiometabolic health. Curr Atheroscler Rep 2026;28:65. doi
  11. Mäkinen KK. Gastrointestinal disturbances associated with the consumption of sugar alcohols with special consideration of xylitol: scientific review and instructions for dentists and other health-care professionals. Int J Dent 2016;2016:5967907. PMC
  12. Oku T, Okazaki M. Laxative threshold of sugar alcohol erythritol in human subjects. Nutr Res 1996;16:577–89. doi
  13. Safety of sugar alcohols on human health: a review. Int J Food Sci Technol 2026;61:vvag012. doi
  14. Zhang R, Noronha JC, Khan TA, et al. The effect of non-nutritive sweetened beverages on postprandial glycemic and endocrine responses: a systematic review and network meta-analysis. Nutrients 2023;15:1050. PMC
  15. Effect of low- and non-calorie sweeteners on the gut microbiota: a review of clinical trials and cross-sectional studies. Nutrition 2023. doi
  16. Witkowski M, Nemet I, Alamri H, et al. The artificial sweetener erythritol and cardiovascular event risk. Nat Med 2023;29:710–18.
  17. Witkowski M, Nemet I, Li XS, et al. Xylitol is prothrombotic and associated with cardiovascular risk. Eur Heart J 2024;45:2439–52.
  18. Witkowski M, Wilcox J, Province V, et al. Ingestion of the non-nutritive sweetener erythritol, but not glucose, enhances platelet reactivity and thrombosis potential in healthy volunteers. Arterioscler Thromb Vasc Biol 2024;44:2136–41. summary
  19. Wölnerhanssen BK, Meyer-Gerspach AC, Arduini A, et al. Sweeteners: erythritol, xylitol and cardiovascular risk—friend or foe? Cardiovasc Res 2025;121:1319–29. doi
  20. Witkowski M, Hazen SL. Erythritol and xylitol and cardiovascular disease risk: a growing concern (reply). Eur Heart J 2025. PMC
  21. Nonnutritive sweeteners and cardiovascular risk: correlation is not causation. JACC Adv 2026;5(3). PMC
  22. Ruskoné-Fourmestraux A, Attar A, Chevalier D, et al. A digestive tolerance study of maltitol after occasional and regular consumption in healthy humans. Eur J Clin Nutr 2003;57:26–30. doi
  23. McCann D, Barrett A, Cooper A, et al. Food additives and hyperactive behaviour in 3-year-old and 8/9-year-old children in the community: a randomised, double-blinded, placebo-controlled trial. Lancet 2007;370:1560–67.
  24. EFSA. Assessment of the results of the study by McCann et al. (2007) on the effect of some colours and sodium benzoate on children's behaviour. EFSA J 2008;660:1–54; and EFSA updates safety advice on six food colours (2009). efsa.europa.eu
  25. FDA. FDA to revoke authorization for the use of Red No. 3 in food and ingested drugs. 15 January 2025. Compliance dates: food 15 January 2027; drugs 18 January 2028.
  26. State food additive & dye bans: 50-state tracker (July 2026) and FDA food-dye phase-out timeline. regbase.com; policycanary.io
  27. Chakravartti SP, Jann K, Veit R, et al. Non-caloric sweetener effects on brain appetite regulation in individuals across varying body weights. Nat Metab 2025;7:574–85 (sucralose, hypothalamic blood flow, hunger).
  28. California OEHHA. Health effects assessment: potential neurobehavioral effects of synthetic food dyes in children. April 2021.
Link copied