Nutrition

What to replace sugar with: the evidence on sweeteners

By Lena Filatova · 9 July 2026 · 7 min read

🇺🇦 Читати українською

When I first wrote about natural sweeteners, the story felt settled: stevia and the sugar alcohols were the "safe" choices, and the main thing to get right was taste. The research has moved on since then — in places, dramatically.

A couple of the sweeteners I once recommended most warmly now come with genuine caution flags. The World Health Organization has weighed in. And a new generation of sweeteners has arrived. So this is a proper rewrite, not a touch-up.

The headline hasn't changed: sugar is worth cutting back on, and no substitute tastes exactly like it. But which substitute, and how much, matters more than "natural versus artificial." Some of the least helpful options are entirely natural — honey and agave among them — and "natural" was never the same thing as "harmless." Let me walk through them the way I'd talk it through with a client.

First, what "safe" actually means here

Almost all of these are safe in the narrow sense that regulators such as the FDA and EFSA permit them and they won't poison you. What we actually care about is subtler: does it spike your blood sugar, does it help or harm weight and metabolic health over years rather than days, and does it carry any newer signal we should respect? That last question is where the picture has changed most.

One honest caveat before we start. Much of the human evidence is still short-term, and some of the worrying new findings are associations plus a plausible mechanism — not proof that the sweetener caused harm. I'll flag where that's the case rather than overstate it.

Stevia — still my everyday recommendation

Stevia, from the leaves of Stevia rebaudiana, remains the one I reach for first. It's calorie-free and doesn't raise blood glucose or insulin ↗. Earlier worries that it disturbs the gut microbiome have not translated into clear harm in humans — a review of the evidence on stevia and gut bacteria ↗ found the human data reassuring rather than alarming, which matters given how much noise there was on this a few years ago.

Its taste has two quirks worth knowing. First, the sweetness comes on slowly — it arrives a moment later than you'd expect — but then lingers pleasantly for a long time. Second, and separately, stevia turns bitter if you use too much: it's intensely sweet, so a heavy hand tips it into a liquorice-like, faintly metallic aftertaste, though used lightly that bitterness simply never appears. That slow-but-lasting sweetness is exactly why stevia pairs so well with erythritol, whose sweetness does the opposite (more on that below). The leaf brewed as tea is the gentlest form; powders, tablets and syrups vary a lot by brand. My practical tip is unchanged — use far less than you think you need, and blend it with something that rounds out the flavour, such as erythritol, monk fruit or a little allulose.

Monk fruit — a genuinely good newer option

Monk fruit (luo han guo) extract wasn't on most people's radar when I first wrote this, and it deserves to be now. It's zero-calorie with a glycaemic index of essentially zero. A 2025 systematic review of randomised trials ↗ found that monk fruit extract doesn't raise blood glucose or insulin, and several trials showed reduced post-meal glucose responses. The evidence on the gut is limited but so far reassuring — no signal of harm. Long-term human data are still thin, so I hold it loosely, but for now it's one of the better choices, especially blended with stevia to soften each one's aftertaste.

Allulose — the newcomer worth knowing

Allulose is the most interesting arrival since the original article. It's a "rare sugar" found naturally in small amounts in figs, raisins and wheat ↗, and it tastes remarkably like sugar — because chemically it nearly is one. The trick is that the body absorbs it but barely metabolises it, so it contributes only about 0.4 calories per gram, against 4 for sugar ↗, with minimal effect on blood glucose.

The evidence is encouraging. A 2024 meta-analysis of randomised trials ↗ found that allulose actually blunted post-meal glucose spikes in people with type 2 diabetes, with little effect on fasting glucose or HbA1c. It holds FDA "generally recognised as safe" status. The practical downsides are cost, patchy availability — allulose is not yet authorised as a novel food in the EU ↗, and the UK's own approval is still pending — and, like all of these, digestive upset if you overdo it. But if you can get it, it's a strong candidate for baking, where stevia and monk fruit struggle.

Erythritol — where I've changed my advice

I used to put erythritol alongside stevia as one of the safest choices. I'd now soften that considerably.

In 2023, a study in Nature Medicine ↗ linked higher blood levels of erythritol to a raised risk of heart attack, stroke and death, and showed a plausible mechanism: erythritol made platelets more prone to clotting. A 2024 follow-up in healthy volunteers ↗ found that a single typical serving increased platelet reactivity even in young, healthy people. This is the same research group that then reported a very similar signal for xylitol.

Two things keep me from panicking, and I'd want you to hold both. First, this is largely observational plus mechanistic; it isn't yet proof that eating erythritol causes heart attacks. Second, the body also makes erythritol from glucose ↗, so high blood levels may partly reflect underlying metabolic problems rather than diet alone. But when a signal appears across multiple study types and points at clotting, the sensible response — especially if you already have diabetes or cardiovascular risk — is to stop treating erythritol as a "use freely" sweetener. Small, occasional use still seems reasonable to me; I no longer suggest it as a daily staple.

On taste, erythritol is stevia's mirror image: its sweetness hits straight away but is "truncated" — it fades almost as fast as it arrives, leaving a slightly cooling, empty finish. That's exactly why the two are so often blended. Erythritol delivers the instant hit of sweetness that stevia lacks; stevia's slow, lingering sweetness carries on after the erythritol has dropped away; and each covers the other's weak spot, with the erythritol masking stevia's bitterness. If you do reach for erythritol, that pairing is where it earns its place.

Xylitol and sorbitol — the same caution, plus the old caveats

Xylitol now carries a twin message. It does have a dental reputation — it's why it's in so much sugar-free gum — though the trial evidence for preventing cavities is weaker than commonly assumed ↗. But a 2024 study in the European Heart Journal ↗ from the same Cleveland Clinic group tied higher blood xylitol to increased cardiovascular events, with the same clotting mechanism seen after ingestion. The caveats above apply here too, but the direction of travel is clear enough that I'd use it sparingly rather than as a table-sugar replacement.

Sorbitol shares xylitol's family and its most reliable feature: in more than modest amounts, both are strongly laxative ↗. That "sugar-free sweets gave me stomach cramps" experience is usually one of these.

One urgent, unrelated warning: xylitol is severely toxic to dogs, even in tiny amounts ↗. If you keep xylitol products at home and have a dog, store them well out of reach.

Isomalt — mostly a confectioner's tool

Isomalt hasn't changed much in the story. It's prized for making clear, hard, decorative caramel that resists melting, so it's a pastry-kitchen material more than an everyday sweetener. Like its sugar-alcohol relatives, it ramps up gut motility and causes bloating and cramping in larger doses. Fine for the occasional showpiece; not something to sweeten your coffee with.

Fructose and the "natural syrups" — natural, but not the win they seem

This is the section I'd underline hardest, because the marketing is so persuasive and the biology so unforgiving.

Fructose isn't really a sugar substitute — it's a sugar. Its low glycaemic index made it fashionable, but the liver handles fructose in a way that, in excess, drives fat production, fatty liver, insulin resistance and raised triglycerides. A 2024 review of the evidence on dietary fructose ↗ has only firmed this up. A little fructose from whole fruit, wrapped in fibre and water, is not the problem; concentrated fructose is.

And that's exactly what the "natural syrups" are. Agave nectar is often held up as a healthy choice, yet it's around 85% fructose ↗ — higher than table sugar — which makes it one of the worse options for metabolic health, not a better one. Honey, maple syrup, date syrup and coconut sugar are all essentially glucose and fructose in prettier packaging. They carry trace minerals and, in honey's case, some antioxidants, so they're not nutritionally identical to white sugar — but for blood-sugar and weight purposes your body treats them much the same. If you're managing diabetes or weight, they don't get a pass for being natural.

The artificial sweeteners — sucralose, aspartame, acesulfame K and saccharin

These four are synthetic rather than plant-derived, and they're what most "diet" and "zero" products actually contain — so they belong in any honest answer to what you'd replace sugar with. The short version: regulators still consider them safe within normal intakes, but the evidence has grown more mixed, and none of them is a reason to relax about sweetness.

Aspartame is the most scrutinised. In 2023 the WHO's cancer agency (IARC) classified aspartame as "possibly carcinogenic to humans" ↗ — its weakest hazard category, based on limited evidence — while the WHO's food-safety committee left the acceptable daily intake unchanged at 40 mg per kg of body weight. That gap matters: "possibly carcinogenic" describes how confident we are that something could cause cancer, not how much risk it poses at real doses. In practice a 70 kg adult would need to drink well over a dozen cans of diet soft drink a day to exceed the limit. People with the rare genetic condition PKU must avoid aspartame ↗; for everyone else, the honest summary is "probably fine in moderation, but not something to drink by the litre." It also loses its sweetness when heated, so it's poor for baking.

Sucralose was long treated as inert, but a 2023 study complicated that. Researchers found that sucralose-6-acetate — an impurity in commercial sucralose, and a compound the gut also forms from it — damaged DNA in laboratory cells ↗, and that a single sucralose-sweetened drink could exceed the safety threshold for genotoxic substances. This is lab work, not proof of harm in people, and regulators haven't changed their stance — but it's a reason not to treat sucralose as consequence-free. Heating sucralose can also generate potentially harmful compounds ↗, so it's best kept out of the oven.

Acesulfame K (often labelled Ace-K) almost always appears blended with aspartame or sucralose to round out the taste. It has been approved for decades and passes through the body largely unchanged. The main open question is the gut: a 2022 trial in Cell found that some non-sugar sweeteners altered the gut microbiome and glucose responses ↗ in some people, with effects that varied from person to person. Acesulfame K itself is less studied here, which is its own argument for modest use rather than confidence.

Saccharin is the oldest of the group. Its 1970s link to bladder cancer in rats was later found not to apply to humans, and it was removed from cancer-warning lists ↗ — so on the classic cancer question it has largely been cleared. The newer question is again the microbiome, where saccharin was one of the sweeteners that shifted gut bacteria and glucose tolerance in the 2022 trial above. It's intensely sweet with a metallic edge, which is why it's usually blended rather than used alone.

My overall take on this group is the same as for the sugar alcohols: not alarming, not a free pass. If you use them, use them as an occasional bridge away from sugar rather than a licence to keep everything sweet.

Where you'll find each one

Sweeteners rarely announce themselves on the front of a pack — they hide in the ingredients list, often two or three blended together. Here's a quick guide to what's typically where, so you can read a label and know what you're looking at.

The safety rating reflects how suitable each option is as a sugar replacement, on current evidence — not a verdict on general food safety. Green = a solid everyday choice; amber = fine in moderation or occasional use; red = behaves much like sugar for your body.

What the big guidelines now say

Two shifts are worth knowing about. In 2023, the WHO advised against using non-sugar sweeteners for weight control ↗, having concluded they don't help with long-term body fat and may be associated with type 2 diabetes and cardiovascular disease over time. Crucially, that guidance explicitly excludes people who already have diabetes, and it's been debated among scientists — but the direction is a nudge away from treating sweeteners as a free lunch for weight loss.

The American Diabetes Association's 2025 Standards of Care ↗ strike a more pragmatic note for people with diabetes: sweeteners can be a reasonable swap for sugar, in moderation and short term, to cut calories and carbs — as long as you don't compensate by eating more elsewhere. That "don't compensate" clause is the whole game.

So how should you actually choose?

Distilled into practical advice:

Best everyday choices: stevia, monk fruit or allulose — ideally blended, since each covers the others' weak spots. Allulose is the standout for baking if you can find it.

Use sparingly, not daily: erythritol and xylitol. The convenience is real; so is the new caution. Small and occasional, especially with cardiovascular risk or diabetes.

Treat as sugar, because they are: honey, agave, maple, date and coconut syrups, and fructose. Natural, lovely, and metabolically about the same as sugar.

The bigger lever: the goal isn't to sweeten everything with a "safe" powder — it's to become gradually less dependent on sweetness overall. Sweeteners work best as a bridge, not a destination.

And the single most important line, unchanged since I first wrote this: pay attention to your own response. People differ genuinely in taste, tolerance and blood-sugar reaction. Try one thing at a time and let your own body — ideally with a glucose meter or CGM if you have diabetes — cast the deciding vote.

What the research says

The information on this website is educational and is not medical advice. Please consult your doctor if you have any doubts or further questions.