Stevia vs Monk Fruit vs Erythritol: Natural Sweeteners for Beverage Formulation

01 At a glance

Stevia vs Monk Fruit vs Erythritol: Side-by-Side Comparison

Most published comparison tables carry three rows: sweetness multiple, calories and origin. Those rows do not decide anything. The table below adds the two properties that most often determine whether a formula survives scale-up — aqueous solubility and the sweetness level actually achievable.

Read down the columns rather than across the rows. Steviol glycosides and mogrosides sit in the same functional class and solve the same problem — the grade ranges are set out on our stevia extract and monk fruit extract pages. Erythritol solves a different one.

02 Sweetness ceiling

Sweetness Potency: Why Stevia and Monk Fruit Have a Sucrose-Equivalence Ceiling

Every high-intensity sweetener has a concentration–sweetness relationship that is sigmoidal, not linear. Sweetness rises with concentration, then flattens as the sweet taste receptor approaches saturation. A psychophysical study that compared sixteen sweeteners on a general labelled magnitude scale found this sigmoidal behaviour across the whole set, and reported that Reb A and acesulfame-K reached markedly lower peak sweetness intensities than sucrose at the upper concentrations tested.

Two consequences follow, and both are practical.

Potency Is a Function of Concentration

A single multiplication factor only describes one point on the curve. Potency is highest near the recognition threshold and falls as concentration rises. Reb A has a reported potency of 250 at 5% sucrose equivalence, while stevioside reaches only about half that — meaning twice as much stevioside is needed for the same sweetness.

So the arithmetic that buyers reach for first:

assumes a proportionality the receptor does not provide. The same caution applies to the grade comparisons on our Reb A and Reb M pages: relative sweetness figures are screening values, not dosing equations.

There Is a Practical Sweetness Ceiling

This is the more serious constraint. Under EU maximum permitted levels, the highest sweetness achievable in a soft drink using pure Reb A is approximately 5–6% sucrose equivalence. A conventional full-sugar carbonated soft drink sits at around 10% sucrose.

The same ceiling logic applies to mogrosides. This is why blending is not a cost-optimisation step in beverage work; it is a structural requirement, and it is the reason the rest of this article is organised around systems rather than single ingredients.

Erythritol is bounded differently, and its own dose–response is not linear either, so the level required to match a given sucrose target cannot be obtained by dividing one number by another. What limits erythritol in practice is not the receptor but two external ceilings: gastrointestinal tolerance per serving, discussed later in this guide, and in the European Union a legal maximum in flavoured drinks that sits well below bulk-sweetener levels.

03 Sensory

Taste and Aftertaste: Stevia vs Monk Fruit Sensory Profiles Compared

At iso-sweetness, these ingredients differ mainly in when sweetness arrives and how long it stays. Time–intensity (TI) sensory work makes this measurable rather than anecdotal.

What Time–Intensity Data Shows for Steviol Glycosides

A TI study of six commonly available steviol glycosides — rubusoside, stevioside, Reb C, Reb A, Reb D and Reb M — found that all parameters extracted from the TI curves varied significantly across the six samples, for both sweetness and bitterness. Reb M and Reb D showed faster sweetness onset, quicker decay of aftertaste and were nearly devoid of bitterness, while rubusoside and stevioside produced an immediate distinct bitter taste and a lingering aftertaste.

More usefully for a purchaser, the study tied this to structure. Fewer glucosyl groups at C-19 were associated with a shorter time to initial stimulation and a longer perception of bitterness, whereas more glucosyl groups at C-13 triggered a faster and stronger rise in sweetness. That gives a mechanistic reason why moving up the glycoside ladder changes the curve, rather than a marketing assertion.

The Finding That Complicates the Upgrade Story

A consumer study with 126 participants compared 0.1% Reb A, Reb D and Reb M against 14% w/v sucrose, measuring in-mouth, immediate and lingering sweetness and bitterness. In-mouth sweetness and bitterness for Reb D and Reb M were not significantly different from sucrose, while Reb A showed significant bitterness — which is the result most suppliers quote.

The part that is quoted less often:

Upgrading from Reb A to Reb M addresses bitterness. On this evidence it does not address sweetness linger, and in this dataset linger was measurably stronger. For a premium beverage where the finish matters as much as the first sip, that is a different formulation problem requiring a different tool. Our stevia aftertaste guide separates the individual off-notes and the variables that move each one.

Where Monk Fruit Actually Sits

Monk fruit is routinely marketed as the cleaner of the two high-intensity options. Receptor-level work complicates that. In a cell-based assay of nine sweeteners against the twenty-five human bitter taste receptors, mogroside V activated at least two TAS2R bitter receptors. Sucralose, acesulfame-K and Reb A showed lower EC₅₀ values for the sweet receptor than for the bitter receptors, which may explain their absence of bitter off-taste at low concentrations — but mogroside V and D-allulose did not share that margin.

This is an in vitro receptor result rather than a sensory panel finding, so it should be read as a mechanistic signal rather than a prediction of what a taster will report. The practical reading is that monk fruit is not inherently bitter-free; its bitterness potential sits at a different point on the concentration curve than stevia's. In a lightly flavoured drink at low use level this rarely shows. In a high-sweetness system it may. If you are choosing between the two categories, Monk Fruit vs Stevia works through the decision in more depth.

Erythritol Is a Sensory Tool, Not Just a Filler

Erythritol has a clean, sucrose-like taste with no reported bitter finish, and is widely used to mask the lingering aftertaste of high-intensity sweeteners that supply sweetness without bulk. The EU legislature accepted exactly this rationale when authorising its use in flavoured drinks: at low levels it acts as a flavour enhancer, helping to mitigate the off-tastes and lingering sweetness associated with high-intensity sweeteners.

04 Solubility

Solubility of Stevia, Monk Fruit and Erythritol in Beverage Systems

This is the single most common way a sweetener that passed bench testing fails in the plant, and it is almost never covered in comparison articles.

Aqueous solubility varies enormously across the steviol glycoside family, and it does not correlate with sensory quality. The better-tasting minor glycosides are frequently the least soluble. The figures below come from different classes of source, and that matters when you are writing a specification, so the source type is shown alongside each number.

Two of these four figures originate in patent applications filed by sweetener producers, measured under conditions the applicant defined. They are widely cited across the industry and are useful as orders of magnitude, but they are not independent peer-reviewed determinations and should not be transcribed into a purchase specification without your own confirmation.

The Syrup Arithmetic

Much of the beverage industry produces carbonated drinks by making a concentrated syrup and diluting it at the filling site, commonly at a ratio of one part syrup to five parts water. For any ingredient to survive that route, its solubility must be at least six times its intended finished concentration.

Apply that to Reb D. Taking the upper end of the reported range, about 450 ppm, the maximum finished concentration achievable through a 1:5 syrup is:

For most beverages that concentration does not deliver meaningful sweetness, and at the lower end of the reported range the figure falls to around 50 ppm. Reported handling data underlines the problem: complete dissolution of Reb D at 0.8% has required heating to near-boiling for around two hours.

None of this means Reb D is unusable. It means Reb D is a blend component and a formulation-design decision, not a drop-in replacement approved on taste alone.

Where Erythritol Changes the Picture

There is a documented physical interaction here that is directly relevant to blended systems. Using a solid dispersion technique, a published study reported an erythritol/Reb D system reaching an aqueous Reb D concentration of 3,000–4,500 mg/L — a five- to six-fold increase over saturation, with an improved dissolution rate as well.

So erythritol is not only contributing bulk and masking off-notes. In that system it also functioned as a solubility enhancer for one of the hardest glycosides to use. That is a stronger technical argument for a blended sweetener system than any of the usual cost reasoning, though it is a laboratory solid-dispersion result and does not automatically transfer to a dry-blended powder. For the underlying physical data across grades, see Reb A vs Reb D vs Reb M Solubility.

05 Stability

Stability in Acidic Beverages: Steviol Glycosides vs Mogroside V

Steviol glycosides are glycosides, so they are subject to acid-catalysed hydrolysis. Across the neutral range they hold up well — reported as quite stable between pH 4 and 8, and tolerant even of UHT processing. Acidic beverages are where the behaviour changes.

Steviol Glycosides Under Beverage Conditions

Reb A has been evaluated in mock beverage solutions simulating cola (pH 2.8 and 3.2), lemon–lime (pH 3.8) and root beer (pH 4.2), at 5, 20, 30 and 40 °C across 26 weeks. Degradation proved dependent on pH, temperature and time together, with the rate of degradation-product formation rising as pH fell and temperature increased, and most degradation product appearing after extended storage. Six minor degradation compounds were structurally characterised. Reported losses reached up to 61% of Reb A at 40 °C after 26 weeks.

Separate work on soft drinks held at 80 °C for 24 to 72 hours recorded degradation of up to 70%, and found stevioside less stable than Reb A. Photostability in beverages has also been studied specifically, so light exposure belongs in any shelf-life protocol alongside pH and temperature.

Reb M has been examined under the same mock-beverage design across pH 2.8–4.2 and 5–40 °C for 26 weeks, with photostability assessed at pH 3.8 under fluorescent light, and its acid degradation products have been isolated and characterised. Newer biocatalytic Reb M analogues developed specifically to address this reported roughly three-fold greater storage stability than Reb M in a commercial low-pH beverage, which indicates the industry treats low-pH stability as an unsolved problem rather than a settled one.

Mogroside V Under the Same Kind of Test

Mogroside V has been evaluated in citric-acidified model systems at pH 3.5 and 5.0, after two different heat-treatment regimens and across 90 days of shelf-life storage, with pasteurised samples and commercial products compared by sensory evaluation. The reported result was that mogroside V remained generally stable across the conditions tested — product pH, both thermal processing regimens, and storage time.

That result is worth stating, and so is its evidential weight relative to the stevia data above it. The steviol glycoside stability figures in this section come from peer-reviewed papers in the Journal of Agricultural and Food Chemistry and Food Research International, with characterised degradation products. The mogroside V comparison rests on a single study published in a horticultural society's conference proceedings, and we are not aware of an equivalent multi-temperature, multi-pH, degradation-product-characterised dataset for mogrosides.

So the honest reading is narrower than “monk fruit is more stable.” It is that mogroside V has not been shown to suffer the acid degradation that steviol glycosides demonstrably do, but the evidence base behind that statement is thinner. If low-pH shelf life is decisive for your product, this is a question to settle with your own accelerated storage trial rather than from either literature, and monk fruit grades and mogroside profiles are set out on our monk fruit extract page.

Erythritol is a different case entirely. It has no reducing end groups, giving it good heat and acid stability with no Maillard browning, and it is practically non-hygroscopic — absorbed moisture stays below 0.5% even at up to 90% relative humidity. It is chemically robust under beverage conditions, which is why the constraints on erythritol in this article are regulatory and physiological rather than chemical.

06 Erythritol's role

What Erythritol Does in a Beverage: Bulk, Mouthfeel and Cooling Effect

Describing erythritol as “a sweetener” understates what it contributes and overstates its sweetening power. At 60–70% the sweetness of sucrose it is a weak sweetener. Its value is elsewhere.

• Body and mouthfeel. Diet beverages built on high-intensity sweeteners alone lack the palate weight consumers associate with sugar. Erythritol restores part of that.
• Off-note suppression. It is used in combination with high-intensity sweeteners to inhibit their undesirable flavours — the basis on which the EU authorised it in flavoured drinks.
• Solubility support in specific blended systems, as described in the previous section.
• A cooling sensation, which is a design variable rather than a defect.

The Physical Properties Behind That

Erythritol is a four-carbon polyol, crystalline and anhydrous, melting at 122 °C. It differs from most simple polyols in having a relatively low solubility of 37% w/w at 25 °C, a property it shares with mannitol and isomalt. It has no reducing end groups, so it does not brown via the Maillard reaction and has good heat and acid stability.

The cooling effect comes from a negative heat of solution. Published values vary with the units and method used, which is worth knowing if you are comparing supplier datasheets: a widely cited figure is −24.1 kJ/mol, equivalent to roughly −180 to −200 J/g at erythritol's molecular weight of 122.12, and a food science reference work gives −181 J/g. Considerably lower per-gram values circulate in secondary sources and do not reconcile with the molar figure, so treat any single number you are quoted as requiring its own citation.

The Cooling Effect Needs a Formulation Decision, Not a Disclaimer

The threshold matters more than the thermodynamic value. Patent literature from a polyol producer reports that aqueous erythritol solutions containing more than 7% w/v are perceived as significantly cooler in the mouth than a sucrose solution of equivalent sweetness at the same temperature. The same source notes the effect cannot be attributed to heat of solution alone, since the erythritol is already dissolved. This is applicant data rather than an independent sensory study, so treat 7% as an approximate design landmark rather than a validated threshold.

Either way, the cooling effect is a matter of matching it to the flavour direction:

Because erythritol and the high-intensity sweeteners are doing genuinely different jobs, comparing them head-to-head on sweetness alone produces a meaningless answer. Stevia vs Erythritol treats them as complementary functions rather than alternatives.

07 Blending

Blending Stevia, Monk Fruit and Erythritol: Which Synergies Are Real?

“Synergy” is used loosely across the sweetener market. Receptor-level work now allows some of those claims to be examined directly.

A 2024 study in Chemical Senses used functional expression of the human sweet taste receptor (TAS1R2/TAS1R3) in a cell-based assay to test nine sweeteners and six binary blends. It confirmed three synergies previously determined by sensory analysis — sucralose with acesulfame-K, rebaudioside A with erythritol, and rebaudioside A with thaumatin — and revealed two new ones, neotame with D-allulose and mogroside V with thaumatin.

It also produced a negative result that deserves more attention than it has received:

Two qualifications belong with that finding, and both point the same way as our commercial interest, so they are worth stating plainly. First, this is an in vitro receptor assay on one binary pair, not a sensory panel: it shows no receptor-level sweetness enhancement, which is not the same as showing that such blends perform no better in a finished beverage. Second, one of the five authors is affiliated with Cargill, a major commercial producer of both erythritol and stevia ingredients — the study confirmed a synergy involving erythritol and found none for the blend that competes with it. Neither point invalidates the work, which is open access with its methods stated, but both belong in front of the reader.

What survives those qualifications is still useful. Stevia-plus-monk-fruit is one of the most heavily promoted combinations in the natural sweetener category, and the specific claim that the two enhance each other's sweetness has, on the best available receptor evidence, no support. There may be sensory, regulatory or supply reasons to combine them. A sweetness-synergy factor should not be assumed in the dosing calculation.

Erythritol Blending Is Not Uniformly Positive Either

A 2026 review reports blend synergy factors in the range of 1.5–2.5× and notes that the effect is strongly ingredient-dependent. Specifically, maltitol suppresses the bitterness and astringency of Reb A, whereas erythritol — despite generally improving temporal sweetness and reducing aftertaste — may enhance bitterness at certain concentrations.

That is an important caution for anyone assuming more erythritol always means a cleaner profile. The relationship is not monotonic, and the concentration at which it inverts is formula-specific. It has to be found by iso-sweet sensory testing across a concentration series, not assumed from a starting ratio. The same logic underlies the function-by-function treatment in Stevia vs Erythritol.

A More Defensible Way to Design the System

• Decide the sweetness target in %SE first, then check whether any single high-intensity sweetener can reach it under your market's legal maxima. Above roughly 6% SE, plan for a system.
• Choose the high-intensity component on the time curve you need, not on the assay number. Bitterness and linger are separate defects with separate solutions.
• Treat erythritol as a dose–response variable, testing a series rather than a single level, because both the cooling threshold and the possible bitterness enhancement are concentration-dependent.
• Validate the blend ratio in the real matrix. Acid system, flavour loading, protein, fat and serving temperature all change the perception.

08 Regulatory

Regulatory Status of Stevia, Monk Fruit and Erythritol by Market

This is where the three ingredients diverge most sharply, and where a formula developed without checking can become unsellable. The position below reflects publicly available information at the time of writing and should be verified against current legislation for your destination market before commercial launch.

Steviol Glycosides

The Joint FAO/WHO Expert Committee on Food Additives established an ADI of 0–4 mg/kg body weight per day, expressed as steviol equivalents, at its 69th meeting, and has reaffirmed it repeatedly since — most recently at the 102nd meeting in June 2026. EFSA and FSANZ set the same figure, and steviol glycosides hold GRAS status in the United States.

Because all steviol glycosides are ultimately metabolised to steviol, exposure is normalised to steviol equivalents using molecular weight ratios. For Reb A the conversion factor is 0.33, derived from the approximate molecular weight ratio of 318:967. Any exposure calculation that skips this conversion will misstate the result.

EFSA has recently been restrictive rather than permissive here. An application to raise the ADI from 4 mg/kg bw/day to either 6 or 16 was assessed and rejected: the Panel concluded there was insufficient justification in the available absorption, distribution, metabolism and excretion dataset, and that the proposed extension of use and increased maximum permitted levels would result in greater ADI exceedance for toddlers at the 95th percentile. Separately, JECFA has confirmed that rebaudioside M and rebaudioside A produced by fermentation fall within the existing ADI, so production route affects the regulatory entry rather than the intake limit. The E 960a–d designations and route distinctions are set out in our Steviol Glycoside Regulation Guide.

Monk Fruit

Monk fruit extract holds GRAS status in the United States and is authorised as a food additive in China under GB 2760. The European position is substantially narrower. Under the Novel Food Regulation (EU) 2015/2283, only one specific aqueous extract has been authorised, from October 2024, under Regulation (EU) 2024/2345. Highly purified mogrosides and non-aqueous extracts remain unapproved, attributed to gaps in toxicological data and the absence of industry-led applications. Certain aqueous decoctions have been recognised as non-novel in the UK and Ireland on the basis of significant consumption before 1997. An earlier EFSA opinion concluded that the toxicity database on monk fruit extract was insufficient to conclude on its safety as a food additive, noting in particular that testicular effects seen in a 90-day study could not be dismissed.

The most recent development is significant enough to change sourcing conversations:

Two things are worth saying about this rather than around it. First, the 38 mg/kg figure comes from a conservative model assuming the proposed maximum levels across the relevant categories, not from measured consumption of existing products; JECFA applies the same conservative approach to steviol glycosides, where predicted maximum exposure for high-consuming young children has similarly been reported at up to 110% of the ADI without the Committee concluding a health concern. Second, a temporary ADI signals an incomplete dossier, not a finding of harm.

It does, however, mean that anyone building a high-sweetness product on monk fruit as the sole high-intensity sweetener is carrying regulatory risk that a blended system reduces, because a blend lowers the required contribution from any single component. The category-level trade-offs are compared in Monk Fruit vs Stevia.

Erythritol

JECFA assigned erythritol an ADI “not specified” in 2000. That position changed in Europe. In December 2023 the EFSA Panel on Food Additives and Flavourings completed a re-evaluation and, in what the Panel acknowledged was the first time an ADI has been derived for a food additive on the basis of an immediate adverse effect, set a numerical ADI of 0.5 g/kg bw/day, using diarrhoea as the critical endpoint. The Panel concluded that both acute and chronic dietary exposure estimates were above that ADI, that individuals with high intake may be at risk of adverse effects after single or repeated exposure, and that the existing laxative warning remains valid. An application to exempt erythritol from that warning label was not supported.

For beverage developers there is a second and more immediate constraint that is widely missed:

A formula designed in a market where erythritol can be used at bulk-sweetener levels will not transfer into that EU beverage category unchanged. At 1.6% in a 330 mL can, the erythritol contribution is roughly 5.3 g, which is a flavour-modifying quantity rather than a sweetening one. If the EU is a target market, that number should be fixed at the start of development, not discovered during registration.

09 Safety literature

Is Erythritol Safe? The Cardiovascular Evidence and Its Limits

Since 2023 this has been the most consequential development in the polyol category, and customers will ask about it. The responsible approach is to describe both the findings and their limitations accurately.

What Was Found

A 2023 study in Nature Medicine examined circulating erythritol and atherothrombotic disease risk. Untargeted metabolomics in a discovery cohort of patients undergoing cardiac risk assessment associated polyol sweeteners, erythritol especially, with three-year incident risk of major adverse cardiovascular events. Targeted analyses in independent US (n = 2,149) and European (n = 833) validation cohorts of stable patients undergoing elective cardiac evaluation reported fourth-versus-first-quartile adjusted hazard ratios of 1.80 (95% CI 1.18–2.77) and 2.21 (95% CI 1.20–4.07) respectively. Mechanistic work showed enhanced platelet reactivity in vitro and thrombosis formation in vivo, and a pilot intervention in healthy volunteers produced marked, sustained elevations in plasma erythritol lasting more than two days.

A follow-up prospective intervention study published in 2024 tested this more directly. Ingestion of 30 g of erythritol, but not 30 g of glucose, produced a greater than 1,000-fold increase in plasma erythritol concentration (6,480 versus 3.75 μmol/L) and acutely enhanced stimulus-dependent platelet aggregation across all subjects, agonists and doses examined, along with increased release of the dense granule marker serotonin and the α-granule marker CXCL4. The authors concluded that discussion of whether erythritol should be re-evaluated as a GRAS food additive is warranted.

What the Counter-Arguments Are

Erythritol is not only a dietary ingredient. It is produced endogenously in humans from glucose via the pentose phosphate pathway, and its synthesis has been shown to rise under oxidative stress. Commentary in the nutrition literature has argued on that basis that elevated plasma erythritol may function as a marker of metabolic dysregulation rather than as its cause, and that observational data cannot distinguish between the two.

A second critique notes that the mechanistic studies demonstrated that erythritol promotes platelet activation, but did not demonstrate that dietary consumption of erythritol produces cardiovascular events through the same pathway. The US FDA has published its own evaluation memorandum summarising the study, placing it against the prior toxicological record and identifying open questions for future work.

The honest summary is that the association is replicated and the acute mechanistic effect is now shown in a controlled human intervention, while causation for clinical events remains unestablished and is contested by researchers who point to endogenous production. That is a more uncomfortable position than either side's headline, and it is where the evidence currently sits.

The Number That Is More Useful in Formulation

Regardless of how the cardiovascular question resolves, gastrointestinal tolerance is the constraint that bites first in beverages, and the data here is specific.

Converting that into beverage terms: a 330 mL can formulated at 6% w/v contains roughly 20 g of erythritol, which sits just above the paediatric NOEL from the study above. A 500 mL bottle at the same concentration contains about 30 g — the dose used in the platelet intervention study. Those are ordinary formulation levels, not extreme ones, which is why the EFSA exposure conclusion and the clinical literature point in the same direction: in beverages specifically, erythritol dose per serving deserves deliberate design rather than being set by sweetness arithmetic alone.

10 By category

Which Natural Sweetener for Which Beverage Category?

The table below is a screening shortlist for pilot work. It gives direction rather than use levels, because iso-sweetness has to be established in your actual matrix.

One pattern runs through all of them. The category names the constraint; the constraint names the candidate; testing decides whether the candidate was right. Where the choice comes down to the two high-intensity categories rather than the grade within one, Monk Fruit vs Stevia sets out the decision in more detail.

11 Sourcing

Sourcing Stevia, Monk Fruit and Erythritol: Supplier Questions That Matter

A specification that carries only an assay number leaves most of the decisions in this article unanswered. These are the questions that separate a specification-based supplier from a catalogue seller.

Enzymatically converted Reb D and Reb M deserve one additional question. Where conversion uses recombinant enzymes, ask whether each batch is tested for residual recombinant DNA and residual protein, and ask to see the method. That is a documentation question a capable manufacturer can answer immediately, and it sits alongside the production-route classifications in our Steviol Glycoside Regulation Guide.

12 Bottom line

Choosing a Natural Sweetener System: The Core Principle

There is no ranking among these three, because they are not doing the same work.

Steviol glycosides give the widest grade range and the most mature regulatory framework, with a clear sensory progression from Reb A through Reb D to Reb M — but a bitterness-versus-linger trade-off that does not disappear at the top of the range, and solubility behaviour that varies by more than an order of magnitude across grades. The full grade range is set out on our stevia extract page, with individual pages for Reb A, Reb D and Reb M.

Monk fruit has a well-liked flavour association and no demonstrated acid-degradation problem in the published model-beverage work, though that evidence base is thinner than the equivalent stevia literature. Against that it has the narrowest European regulatory position and a newly issued temporary ADI with an open data request attached. Grades and mogroside profiles are covered on our monk fruit extract page.

Erythritol is the only one of the three that supplies bulk, and it does real sensory work suppressing off-notes and, in at least one published system, supporting solubility — but it carries the tightest legal constraint in EU beverages, an EFSA ADI that current exposure already exceeds, a cooling threshold that conflicts with some flavour directions, and an unresolved cardiovascular literature that customers will ask about.

Which leaves a single operating rule:

If you are screening options, send us your beverage category, target sucrose equivalence, finished pH, thermal process, destination market and highest in-process sweetener concentration. We can propose a starting system for pilot testing and supply the relevant specifications, COAs and samples — request a quotation or free sample here.

13 Questions

Frequently Asked Questions

14 Evidence base

References

Sources are grouped by subject. Where a figure comes from patent literature, conference proceedings or an industry-affiliated study, that is noted in the entry.

Sensory and sweetness behaviour

[1] Dynamic characteristics of sweetness and bitterness and their correlation with chemical structures for six steviol glycosides. Food Research International. 2021. Article identifier PII S0963996921007481. Time–intensity characterisation of rubusoside, stevioside, Reb C, Reb A, Reb D and Reb M, including the C-13 and C-19 glucosylation effects on temporal sweetness and bitterness. ScienceDirect

[2] Tao R, Cho S. Consumer-Based Sensory Characterization of Steviol Glycosides (Rebaudioside A, D, and M). Foods. 2020;9(8):1026. DOI: 10.3390/foods9081026. Source of the lingering-sweetness finding for Reb D and Reb M against a 14% w/v sucrose reference, using a 126-participant consumer panel and a CATA question. PMID 32751835

[3] A Comparison of Psychophysical Dose-Response Behaviour across 16 Sweeteners. Nutrients. 2018;10(11):1632. DOI: 10.3390/nu10111632. Establishes sigmoidal dose–response across nutritive, polyol and non-nutritive sweeteners including Reb A and mogroside V. PMC6266678

[4] Institute of Food Science and Technology. Stevia: a non-caloric sweetener of natural origin. Information Statement. Source of the 5–6% sucrose equivalence ceiling under EU limits and the concentration-dependence of potency. IFST

[5] Stevia-derived sweeteners in reduced-sugar foods: formulation strategies, sensory optimization, nutritional evidence, and safety perspectives. Frontiers in Nutrition. 2026. DOI: 10.3389/fnut.2026.1894194. Narrative review of the bitterness, linger, licorice character, thin mouthfeel and matrix dependence that limit steviol glycoside application. Article

Receptor-level and blending studies

[6] Belloir C, Jeannin M, Karolkowski A, Scott C, Briand L. A receptor-based assay to study the sweet and bitter tastes of sweeteners and binary sweet blends: the SWEET project. Chemical Senses. 2024;49:bjae041. DOI: 10.1093/chemse/bjae041. Open access. Cell-based TAS1R2/TAS1R3 and TAS2R screening of 9 sweeteners and 6 binary blends; source of the confirmed Reb A/erythritol synergy, the absence of synergy for the Reb M/mogroside V blend, and the TAS2R activation by mogroside V. Conflict-of-interest note: four authors are at the Centre des Sciences du Goût et de l'Alimentation, Dijon; one author is affiliated with Cargill Core Research and Development, a commercial producer of both erythritol and stevia ingredients. Oxford Academic

[7] A Comprehensive Review on Steviol Glycosides: Sources, Properties, Bioactivities, Sensory-Functional Enhancement and Bioproduction Strategies. Plants. 2026;15(2):324. Reports blend synergy factors of 1.5–2.5× and the ingredient-dependent behaviour of maltitol versus erythritol on Reb A bitterness. MDPI

Solubility and physical behaviour of steviol glycosides

[8] Characterization of a new hemihydrate rebaudioside B crystal having lower aqueous solubility. Food Chemistry. 2019. Comparative room-temperature aqueous solubility figures for Reb B, Reb D and Reb M. Note that the comparative values are reported as in-house determinations within a peer-reviewed paper. ScienceDirect

[9] Investigation of the solubility enhancement mechanism of rebaudioside D using a solid dispersion technique with potassium sorbate as a carrier. Food Chemistry. 2015. Includes the erythritol/Reb D system reaching 3,000–4,500 mg/L, a five- to six-fold increase over saturation, under laboratory solid-dispersion conditions. ScienceDirect

[10] Patent literature, sweetener manufacturers. The Reb A (≥3,000 ppm) and Reb D (300–450 ppm) room-temperature solubility figures cited in this article, and the 1:5 syrup dilution rule requiring roughly six times the finished concentration, originate in applicant-supplied data within sweetener patent filings. They are widely reproduced in industry technical literature but are not independent peer-reviewed determinations and should be confirmed against your own material.

[11] Dual-Enzyme Biocatalytic Synthesis of Novel Rebaudioside M Analogs with Enhanced Aqueous Solubility, Physicochemical Stability, and Sensory Profiles. Reports roughly three-fold greater storage stability than Reb M in a commercial low-pH beverage. PMC13464401

Stability in beverage systems

[12] Prakash I, Clos JF, Chaturvedula VSP. Stability of rebaudioside A under acidic conditions and its degradation products. Food Research International. 2012;48(1). DOI: 10.1016/j.foodres.2012.02.015. Mock beverages at pH 2.8–4.2 and 5–40 °C over 26 weeks, with six degradation products characterised. Authors were affiliated with a beverage manufacturer at the time of publication. ScienceDirect

[13] Wölwer-Rieck U, Tomberg W, Wawrzun A. Investigations on the stability of stevioside and rebaudioside A in soft drinks. Journal of Agricultural and Food Chemistry. 2010;58(23):12216–12220. DOI: 10.1021/jf102894v. Degradation of up to 70% at 80 °C, with stevioside less stable than Reb A. ACS

[14] Clos JF, DuBois GE, Prakash I. Photostability of rebaudioside A and stevioside in beverages. Journal of Agricultural and Food Chemistry. 2008;56(18):8507–8513. DOI: 10.1021/jf801343e. ACS

[15] The use and stability of monk fruit plant-derived sweetener in orange juice beverages. Proceedings of the Florida State Horticultural Society. Mogroside V in citric-acidified model systems at pH 3.5 and 5.0 across two heat treatments and 90 days of storage. Evidence note: this is a conference proceedings paper, not a peer-reviewed journal article, and it is the single source for the mogroside V stability statement in this guide. It is not equivalent in weight to references [12]–[14]. Full text

Erythritol physicochemical properties

[16] Erythritol — an overview. ScienceDirect Topics, compiled from food science and pharmaceutical reference works. Source of the 37% w/w solubility at 25 °C, the 122 °C melting point, the absence of reducing end groups and consequent heat and acid stability, and a heat of solution of −181 J/g. ScienceDirect Topics

[17] Perko R, DeCock P. Erythritol. In: Optimising Sweet Taste in Foods. Woodhead Publishing. Source of the negative heat of solution of −24.1 kJ/mol and the description of the cooling effect as desirable in mint and menthol products but objectionable in others. Chapter record

[18] Erythritol: A New Multipurpose Excipient. Pharmaceutical Technology. Source of the non-hygroscopicity data: practically non-hygroscopic at up to 90% relative humidity with absorbed moisture below 0.5%. Evidence note: this trade-press article also reports a heat of solution of 243 cal/g, which does not reconcile with the molar value in reference [17]; the molar figure is used in this guide. Article

[19] US Patent 6,066,345 (Cerestar Holding B.V.). Erythritol containing beverage imparting a cooling taste sensation. Source of the finding that aqueous erythritol solutions above 7% w/v are perceived as significantly cooler than an iso-sweet sucrose solution at the same temperature. Evidence note: patent applicant data, not an independent sensory study. Google Patents

Regulatory

[20] Joint FAO/WHO Expert Committee on Food Additives. Summary and conclusions, 102nd meeting, Nanjing, 9–18 June 2026. Temporary ADI of 0–10 mg/kg bw/day for monk fruit extract expressed as mogroside V, confirmation of the steviol glycoside ADI at 0–4 mg/kg bw/day, and the children's exposure estimate of 38 mg/kg bw/day. Evidence note: the figures quoted in this guide were taken from reporting on the published summary rather than from the primary document; verify against the WHO original before regulatory use. JECFA

[21] WHO JECFA food additive database — steviol glycosides. Record of the 0–4 mg/kg bw ADI expressed as steviol, its NOEL basis, the inclusion of fermentation-derived rebaudioside M and rebaudioside A, and the predicted maximum exposure of 4.0–4.4 mg/kg bw per day for high-consuming young children. JECFA database

[22] EFSA Panel on Food Additives and Flavourings. Re-evaluation of erythritol (E 968) as a food additive. EFSA Journal. 2023;21(12):e8430. DOI: 10.2903/j.efsa.2023.8430. Establishes the 0.5 g/kg bw/day ADI based on diarrhoea as the critical endpoint and concludes exposure exceeds it. PMC10731997

[23] Commission Regulation (EU) 2015/1832 of 12 October 2015 amending Annex II to Regulation (EC) No 1333/2008 as regards the use of erythritol (E 968) as a flavour enhancer in energy-reduced or no-added-sugars flavoured drinks. Source of the 1.6% flavour-enhancer limitation in food category 14.1.4. EUR-Lex

[24] Why Does Monk Fruit Extract Remain Only Partially Approved in the EU? Regulatory Barriers and Policy Implications for Food Innovation. Foods. 2025;14(16):2810. DOI: 10.3390/foods14162810. Systematic review of the EU novel food position, including Regulation (EU) 2024/2345 and the UK and Ireland non-novel treatment of aqueous decoctions. PMID 40870722

[25] EFSA Panel on Food Additives and Flavourings. Safety of use of monk fruit extract as a food additive in different food categories. EFSA Journal. 2019. Concludes the toxicity database is insufficient, including the testicular effects observed in a 90-day study that could not be dismissed. PMC7008860

[26] EFSA Panel on Food Additives and Flavourings. Scientific opinion on the extension of the authorisation of use of steviol glycosides (E 960a–d) and the modification of the ADI for steviol. EFSA Journal. 2024. Rejection of the proposed increase from 4 to 6 or 16 mg/kg bw/day and the toddler exposure conclusion. PMC11533382

[27] World Health Organization. Use of non-sugar sweeteners: WHO guideline. Geneva: WHO; 2023. Defines the scope of the recommendation and the exclusion of polyols. WHO

Erythritol safety and tolerance

[28] Witkowski M, Nemet I, Alamri H, et al. The artificial sweetener erythritol and cardiovascular event risk. Nature Medicine. 2023;29:710–718. DOI: 10.1038/s41591-023-02223-9. PMID 36849732

[29] 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. Arteriosclerosis, Thrombosis, and Vascular Biology. 2024;44(9):2136–2141. DOI: 10.1161/ATVBAHA.124.321019. PMID 39114916

[30] Cramer T, Gonder U, Kofler B. Plasma erythritol and cardiovascular risk: is there evidence for an association with dietary intake? Frontiers in Nutrition. 2023;10:1195521. DOI: 10.3389/fnut.2023.1195521. Critical evaluation including the endogenous pentose phosphate pathway argument. PMC10242034

[31] Elevated erythritol: a marker of metabolic dysregulation or contributor to the pathogenesis of cardiometabolic disease? Review noting that the mechanistic data did not demonstrate that dietary erythritol consumption promotes cardiovascular events through the platelet pathway. PMC10534702

[32] US Food and Drug Administration. An evaluation of the article “The artificial sweetener erythritol and cardiovascular event risk.” Agency memorandum summarising strengths, weaknesses and unanswered questions. FDA

[33] Gastrointestinal tolerance of erythritol-containing beverage in young children: a double-blind, randomised controlled trial. European Journal of Clinical Nutrition. 2015. Establishes the 15 g (0.73 g/kg bw) no observed effect level in a 250 mL beverage. Study funded by an erythritol manufacturer. PMID 25744158

[34] Gastrointestinal tolerance of erythritol and xylitol ingested in a liquid. European Journal of Clinical Nutrition. Adult single-bolus tolerance at 20 g, 35 g and 50 g in water. PMID 16988647