If you have just started keto, you have probably noticed that sweeteners come with a lot of confident claims. “Zero carbs.” “Won’t spike your blood sugar.” “Keto friendly.” Some of that is true. Some of it is only half true. And a few products are leaning on the label a bit more than the science would like.
This article is the companion to Keto Sweeteners: What to Use and When, which covers what to bake with and when. This one looks behind the label: what the research says about blood sugar, which sweeteners you can take off your carb count, and what is worth knowing before you start.
If you only read one paragraph, make it this one. Erythritol is the one you can subtract in full. Allulose is the only sweetener with trial evidence that it can lower the blood sugar rise from other carbs eaten with it. And some things that look keto on the packet, like maltitol, still count. Most of the research is small and short, so I have said where it is solid and where it is not.
This is general information, not advice tailored to you. If you have diabetes or take medication, your doctor or dietitian comes first.
What the research shows, and whether to count it
Here is the honest bit: most claims of “no effect on blood sugar” rest on small, short trials in healthy people, and they are often compared against sugar rather than against nothing. For a sense of scale, table sugar scores a glycaemic index (a score for how quickly a food raises blood glucose) of 65 and an insulin index of 43 in one widely cited compilation (Livesey 2003).
| Sweetener | What the research found | Evidence | Count it? |
|---|---|---|---|
| Erythritol | Glycaemic index 0, insulin index 2. 36 g a day for 4 weeks in 42 adults with obesity did not change glucose or insulin (Bordier 2023). | Low to moderate | Subtract in full |
| Allulose | Modestly lowered the glucose response to a carbohydrate meal in pooled trials of healthy adults (Tani 2023). Lower glucose response in type 2 diabetes, but no HbA1c benefit shown. | Best of the group, but short and small | Count as zero, within the gut limit |
| Xylitol | Glycaemic index 13, insulin index 11: small, but not zero. I could not find a trial in people with diabetes. | Low | Subtract only with a stated assumption |
| Monk fruit | Less glucose and insulin than sugar in two small crossovers. That is sugar removed, not glucose lowered (review). | Low | Zero for pure extract. Check blends for fillers |
| Stevia | No significant change in glucose tolerance in one two-week trial in healthy adults (Suez 2022). | Very low | Zero for pure extract. Check blends for fillers |
| Maltitol | Glycaemic index 35, insulin index 27. Fifty grams gave about 23 to 34% of the glucose response of 50 g glucose (Pratt 2011). | Low to moderate | Count mostly or fully |
| Inulin and fructo-oligosaccharides | Lowered the glucose response of a food mainly by replacing part of its sugar (European Journal of Nutrition 2017). | Low | Unresolved. Count if you are tight on carbs |
| Isomalto-oligosaccharide syrups | Raised glucose to 132 mg/dL at 30 minutes in a 9-person trial, versus about 88 for comparison fibre and placebo (Journal of Metabolic Health). | Very low (one small trial) | Count as carbohydrate |
| Sucralose, saccharin, aspartame | Mixed. Saccharin and sucralose worsened glucose tolerance in one small trial and aspartame did not (Suez 2022). A Cochrane review in diabetes found no HbA1c difference, very low certainty (Cochrane). | Low to very low | Zero, but evidence is conflicting |
The last column is my own practical suggestion based on this evidence, not an official standard. Three things worth remembering:
- “Less glucose than sugar” is not “lowers blood sugar”. Say the first for monk fruit and stevia. Keep the second for allulose, and only with other carbohydrate in the same meal.
- People differ a lot. In 1,002 healthy adults, the response to identical meals varied widely between people (PREDICT 1). If you have a glucose meter, using it is sensible, though no trial has tested that approach for sweeteners.
- Ketones are a gap. No controlled human trial measuring blood ketones after these sweeteners was found. A dietitian-supervised epilepsy sheet says erythritol, allulose, stevia and monk fruit “do not tend to affect ketone levels” but cites no studies (Ketogenic Dietitians Research Network 2025). Treat “will not kick you out of ketosis” as likely, not proven.
When to count the carbs, and when not to
“Net carbs” sounds like one simple rule. It is not. It is really three different things that can give three different answers for the same ingredient: the label rule (how the manufacturer must calculate the carbohydrate figure), the science (what the ingredient does to your glucose) and the keto convention (“total carbs minus fibre and sugar alcohols”). The danger sits between the first two: a label can legitimately leave an ingredient out of the carbohydrate figure while it still raises your blood glucose.
Sugar alcohols (polyols) such as erythritol, xylitol, sorbitol and maltitol are only partly absorbed. Monash University puts it at about 30% (Monash). That is why they are lower in energy than sugar, and why they can cause gas and diarrhoea. How much glucose they release varies a lot between them.
Eight rules for beginners
Think of these as a starting point rather than a rulebook. They are my practical rules, not an official standard.
- Start with the per 100 g column and the ingredient list, not the front of pack or a recipe’s own “net carb” figure. Australian panels may print “less than 1 g”, and polyols are often missing from panels outside the USA (Glycemic Index Foundation).
- Subtract erythritol in full.
- Treat allulose as zero carbohydrate, but watch the dose (see safety below).
- Count maltitol and hydrogenated starch hydrolysate mostly or fully. Their glycaemic index values are 35 and 39.
- Do not subtract isomalto-oligosaccharide syrups, maltodextrin or dextrose. Splenda packets list dextrose and maltodextrin first and declare 1 g carbohydrate per 1 g packet.
- Subtract xylitol and sorbitol only with a stated assumption, either the “half” convention (UCSF) or count them.
- Treat resistant dextrin and inulin as unresolved. I have no reliable number, so count them if you are tight on carbs.
- Add everything up across a recipe. Sweeteners and gut limits stack.
Three worked examples
Here is how that looks in a real kitchen. This is illustrative arithmetic, not measured values.
- Erythritol blend, US panel. A retailer lists Lakanto classic as 3 g total carbohydrate and 3 g sugar alcohol per packet, so net is 3 minus 3, which is 0 (ShopRite). US panels typically subtract the whole figure, so compare ingredient lists, not panels, across the two systems.
- Chocolate with 30 g maltitol. The “half” rule counts 15 g. Scaling the Pratt result gives roughly 7 to 10 g. Both are well above zero, and stretching a 50 g trial to a 30 g bar is only an estimate.
- Twelve “zero calorie” sachets. Each declares 1 g carbohydrate, so a recipe using 12 carries about 12 g. Where a panel rounds a small amount down to “0”, the real total could be several grams more.
If you have diabetes, especially if you use insulin or a sulfonylurea, please get advice from your doctor or an accredited practising dietitian and check with a meter. Sources conflict on what the American Diabetes Association (the main US diabetes body) says about subtracting sugar alcohols, and I could not reach its own guidance, so I have not attributed any rule to it.
Safety: gut limits, the heart question and regulators
Nothing here is risk-free, but the good news is that the risks are mostly about your gut. There is also one heart question that researchers are still arguing about.
| Topic | What to know |
|---|---|
| Gut tolerance | Polyols and allulose draw water into the bowel, and doses that suit one person upset another. In single-dose trials, 50 g of erythritol caused nausea and rumbling while 20 and 35 g did not significantly, and 35 g of xylitol caused watery stools (Storey 2007, industry-sponsored). A pack’s laxative advisory is per 100 g of food, not per person, so you can pass your own limit across a day. My advice: start small, change one sweetener at a time and spread them across the day. That is expert opinion, as no trial has tested a beginner build-up. If you have irritable bowel syndrome, Monash notes polyols are a well-documented trigger. |
| Allulose limits | Food Standards Australia New Zealand adopted 0.4 g per kg of body weight per serve and 0.9 g per kg per day, roughly 28 g per serve for a 70 kg adult, based on a small industry-linked Korean trial in which 13 of 29 people had diarrhoea at 0.5 g/kg (A1247 approval report). Treat that as a ceiling, not a starting dose. It also noted there was no assessment specific to children or people with diabetes. |
| Allulose in Australia | Food Standards Australia New Zealand approved it as a novel food on 7 August 2024, with a laxative advisory on many foods, and counts it at 2 kJ/g, excluded from declared sugars. I could not confirm the gazettal date or whether the original exclusive supplier arrangement has lapsed, so check current brands. The European Food Safety Authority’s June 2025 opinion said safety “cannot be established” because the applicant did not supply requested data. That reads as incomplete data, not an identified hazard. |
| Erythritol and the heart | A 2023 Nature Medicine study linked high blood erythritol to about twice the rate of major cardiac events in high-risk patients, and a small 2024 trial of 30 g found increased platelet reactivity (Cleveland Clinic). The US Food and Drug Administration noted association is not causation, that kidney function and diet were not adjusted for, and that most samples pre-dated widespread erythritol use (US Food and Drug Administration memo). The body also makes erythritol, and a 2025 review found no proven causal link. The fair way to put it is a contested association with no outcome trial. It is not a reason to panic, but be a little cautious if you have diabetes, obesity or heart disease. |
| World Health Organization and sweeteners | In 2023 the World Health Organization conditionally advised against non-sugar sweeteners for weight control, including stevia, aspartame and sucralose. It does not cover people with diabetes, nor sugar alcohols and low-calorie sugars such as allulose, erythritol and xylitol, and outside experts rated the evidence low certainty (PAHO/WHO). Its practical advice, to reduce the overall sweetness of your diet, suits keto well. |
Before you start keto
Talk to your doctor or an accredited practising dietitian first if you take insulin or a sulfonylurea (low blood sugar risk), take a sodium-glucose cotransporter 2 inhibitor, a type of diabetes medicine (it can cause ketoacidosis even with normal glucose), have type 1 diabetes or kidney disease, are pregnant, breastfeeding or under 18, or have a current or past eating disorder. Please do not stop any medicine yourself. Australian diabetes bodies describe low-carb as a short-term option for some adults with type 2 diabetes (about 6 months in people who are overweight, per Dietitians Australia). None of them endorses a particular sweetener or a way of subtracting sugar alcohols (Diabetes Australia, Dietitians Australia, Australian Diabetes Educators Association).
Cholesterol and bloods. LDL cholesterol, the type linked to heart disease, rises in some people on keto. A Mayo Clinic series of 17 adults whose LDL cholesterol was 190 mg/dL or higher showed it can climb a long way and fall after stopping (2023). That shows it is possible, not how often. I found no Australian body recommending baseline and follow-up bloods, so my suggestion to check cholesterol, HbA1c (a three-month average of blood glucose) and kidney function before starting and after a few months is reasonable practice, not official guidance.
If you have a dog, xylitol is an emergency. US veterinary sources put low blood sugar risk above about 0.1 g per kg and liver damage risk above about 0.5 g per kg, with symptoms from 10 minutes to 12 hours (ASPCApro). Do not wait for symptoms. Call a vet or the Animal Poisons Helpline on 1300 869 738 (24 hours), and do not induce vomiting unless a vet says to. Xylitol can also appear as “birch sugar”. I found no veterinary information on erythritol, allulose, stevia or monk fruit, so I would not call them safe for dogs either.
That is the lot. None of it is meant to put you off. It is just easier to start well than to untangle a messy first few weeks, and your future self will thank you for checking the label.
