An agar gel sets between 32 and 43 degrees and will not melt again below 60 degrees. That single fact drives everything else. Gelatin melts at roughly 30 to 35 degrees, which is to say in your mouth. Those thirty degrees of difference are why agar is not a drop-in replacement for gelatin: it is a different gel, with different strengths and different disappointments. It holds on a summer table without slumping, it cuts cleanly, and it will never give you the melting texture of a classic panna cotta.

The rest of this page gives the doses in grams per litre with the texture each one produces, an honest conversion from a gelatin recipe, and the three real reasons a preparation fails to set. One of them is almost always blamed on the wrong thing.

What agar-agar actually is, chemically

The FAO/WHO Joint Expert Committee on Food Additives defines agar as the dried hydrophilic colloidal substance extracted from certain marine algae of the class Rhodophyceae, the red seaweeds. It is a polysaccharide, consisting primarily of D- and L-galactose units, with roughly every tenth D-galactopyranose unit carrying a sulfate ester group.

That word, polysaccharide, explains most of what follows. Agar is a chain of sugars. It is not a protein. Gelatin is a protein, produced by partial hydrolysis of collagen. Two different families of molecule, two ways of building a gel, two separate lists of things that can go wrong.

In detail, agar is a mixture of two fractions: agarose, around 70 %, and agaropectin, around 30 %. Agarose is a linear polymer of agarobiose units, themselves made of D-galactose and 3,6-anhydro-L-galactopyranose. Agarose does the gelling. Agaropectin is sulfated, contains D-glucuronic acid and pyruvic acid, and gels poorly. The ratio between them, and above all the degree of sulfation, decides how good a powder is: the more sulfated the agar, the less the long helices associate, the more the hydrogen bonds between molecules are disrupted, and the weaker the gel.

This is also why one packet of agar is not interchangeable with another. Gel strength is officially measured on a 1.5 % gel using the Nikkansui method, and it varies by a factor of two between batches depending on the seaweed species and the extraction process. No consumer brand prints that figure on the sachet. The practical consequence is simple: your first attempt with a new powder is a calibration, not a result.

Temperatures matter more than dosage

Agar is insoluble in cold water and soluble in boiling water. Once dissolved, the solution cools and sets into a gel between 32 and 43 degrees. To melt that gel again, you have to take it above 60 degrees.

The gap between setting and melting has a name: hysteresis. For agar it sits around 40 to 60 degrees, and it is the most useful property of the ingredient. The official identification test in the JECFA monograph describes it precisely: a 1.0 % solution held at 30 degrees for fifteen minutes forms a firm, resistant gel; that gel is still not molten after an hour at 70 degrees; it liquefies into a clear solution only above 95 degrees.

Three consequences for cooking, in order of importance:

An agar dessert can be served at room temperature without worry. A summer terrine, a buffet, a picnic: the gel holds. That is exactly what gelatin cannot do.

In exchange, the gel does not melt in the mouth. It breaks down through chewing and saliva, not through body heat. The sensation is clean, almost brittle, never creamy. A great many disappointments with agar come from this rather than from a dosing error.

Finally, setting is fast and starts early. As soon as the mixture drops below 43 degrees the structure begins to form. Pouring into moulds promptly is not fussiness: a mixture that has begun to set in the pan produces a grainy gel, broken by the act of pouring.

Doses, in grams per litre

Agar has to be weighed. A level teaspoon holds between 2 and 3 g depending on how finely the powder is milled and how you fill the spoon, which is a 50 % swing on the final result. A scale reading in grams, ideally in tenths of a gram, costs less than one wasted batch.

Agar per litre of liquid	Result	Use
1 g	Slightly thickened liquid, will not hold a cut	Glaze, sauce, thickened coulis
2 g	Soft, wobbly gel, hard to unmould	Dessert cream, spoonable flan
3 g	Clean gel, unmoulds, still tender	Plant-based panna cotta, entremets
4 g	Firm gel, cuts cleanly	Fruit jelly, terrine, aspic
6 g and above	Hard, brittle, rubbery in the mouth	Yōkan-style confectionery, cubes

These figures assume a neutral liquid, neither strongly acidic nor heavily sweetened, and a powder of average strength. They are a starting point to adjust, not a law. The progression is not perfectly linear either: between 2 and 4 g the shift from wobbly to firm is sharp, whereas past 6 g the texture mostly gains hardness without gaining much structure.

A published trial on plant-based jellies confirms the same logic at more generous quantities: tested at 4, 8 and 12 g of agar, firmness rose progressively, with the 12 g jellies the firmest of the series.

One precaution deserves its own line. The powder must be dispersed in the cold liquid, whisking, before any heat is applied. Thrown into an already hot liquid it instantly forms gelled lumps on the surface that will never dissolve, and the inside of each lump takes no part in the gel. The dose then looks too low when it was in fact correct.

Replacing gelatin: the honest conversion

A gold-grade gelatin leaf weighs about 2 g. Many recipes count in leaves without ever giving a weight, which is a first source of error before agar even enters the picture.

For a standard moulded dessert, a recipe calling for six leaves per litre, about 12 g, translates to roughly 3 to 4 g of agar per litre. So the ratio is about one to three or one to four by mass. But presenting that as a conversion rule would be misleading, for a reason that has nothing to do with gel strength.

The two gels do not do the same job. Swapping agar into an aerated mousse gives a poor result: agar sets too fast and too decisively for a foam to stabilise, and the airy texture collapses into a dense mass. In a bavarois, the same problem. Conversely, for a fruit jelly, a vegetable terrine or a plant-based pâté, agar does the job better than gelatin, because holding at room temperature is precisely what you want.

The right way to think about it is therefore not to convert a recipe but to choose the gel that matches the result you want. Where the original recipe relies on a melting texture, agar alone will not deliver; it has to be combined with fat or with a rich purée, silken tofu, nut butter, coconut milk, to supply the richness the gel does not provide.

Agar-agar is fully compatible with vegetarian and vegan diets, and with kosher and halal requirements, since it contains no animal component. That is the main reason it appears in so many substitution recipes. For the other common swaps in plant-based baking, the guide to egg substitutes for each type of recipe covers the cases where the problem is binding rather than gelling.

The three real reasons a gel fails

Not enough boiling

Agar only gels once it has genuinely dissolved, which means reaching a boil and holding it. Thirty seconds to a minute of proper simmering, stirring, is enough in a thin liquid. In a thick one, fruit purée or a starch-heavy mixture, heat moves poorly and you need longer, whisking throughout so the bottom does not catch.

This is the most common failure and the easiest to fix: a mixture that never truly boiled will not set, whatever the dose.

Acidity

Here is the point most recipes get backwards. In an acidic medium and under heat, the agarose chain undergoes acid hydrolysis: the bonds between sugar units are cut, the chains shorten, and a shortened polymer gels poorly or not at all. The effect worsens with heating time and with acidity.

The consequence is counterintuitive. Faced with an acidic preparation, lemon juice, red fruit, rhubarb, vinegar, the common reflex is to extend the cooking to “help” it set. That is exactly the wrong move: every extra minute of boiling in an acidic medium degrades a little more of the gelling agent.

The method that works runs the other way. Boil the agar in the neutral part of the preparation, water, a low-acid juice, plant milk, for as long as needed. Then, off the heat and once the temperature has dropped a little, stir in the acidic element quickly. Raising the dose slightly, by around a third, compensates for what is lost anyway.

Dosage

This is the least frequent cause, contrary to expectation, precisely because it is the one everybody watches. When a gel is too soft despite proper boiling and controlled acidity, adding one gram per litre usually settles it. When it is rubbery, you have gone past 5 or 6 g per litre.

Fruit enzymes: clearing up a confusion

One instruction turns up in nearly every agar recipe: pineapple, papaya, mango and kiwi must be cooked first, because their enzymes prevent setting.

That instruction is true for gelatin and false for agar. The enzymes involved, bromelain in pineapple, papain in papaya, actinidin in kiwi, are proteases: they cut the peptide bonds in proteins. Gelatin being a protein, they chop it up and destroy its network, which neatly explains every liquid mousse ever made with fresh pineapple. Agar is a polysaccharide, a chain of sugars with no peptide bond anywhere in it. A protease has literally nothing to cut.

Fresh pineapple does still cause trouble for agar, but for another reason: its pH sits around 3.5, so it is acidic, and the acid hydrolysis described above is what does the damage. The fix is the same as for any acidic fruit, and it has nothing to do with blanching to denature enzymes.

This is not chemist’s pedantry. It changes what you do: there is no point cooking the fruit if the only goal is to neutralise enzymes, whereas you do need to manage acidity and the moment at which it goes in.

What agar gives you, and what it does not

Agar is more than 94 % soluble fibre and does not raise the net calories of a food. At culinary doses, a few grams per litre, its nutritional contribution is negligible in both directions: no calories, and no vitamins or minerals in any useful quantity. It is not a food, it is a texture.

The appetite-suppressant effect often attributed to it deserves a close look, because the numbers exist and they do not say what they are made to say.

A randomised trial published in 2005 in Diabetes, Obesity and Metabolism followed 76 obese patients with impaired glucose tolerance or type 2 diabetes. After a four-week run-in they were assigned to either a conventional diet or the same diet with added agar, for twelve weeks. The agar group lost 2.8 kg on average against 1.3 kg in the control group, a difference of 1.5 kg (p = 0.008), with a BMI drop of 1.1 against 0.5 kg/m² (p = 0.009) and total cholesterol down 7.6 mg/dl against a 2.4 mg/dl rise (p = 0.036). The authors explicitly attribute the result to maintaining a reduced calorie intake.

Two clarifications change how those figures read. First, this is agar eaten as a bulking food, at quantities unrelated to the 3 g that gel a litre of flan. Second, both groups lost weight: agar propped up a calorie restriction, it did not replace one. There is no reason to expect anything of the kind from a gelled dessert.

A better-supported use is digestive. A three-week crossover trial published in 2026 in the Asia Pacific Journal of Clinical Nutrition followed 20 constipated elderly residents of a Vietnamese nursing home. Their usual tapioca-and-gelatin snacks were replaced by snacks in which konjac and agar each supplied about 3 g of fibre. Stool frequency rose from 3.3 to 4.9 times a week, the proportion of normal-form stools from 35 to 85 %, and the constipated form fell from 56 to 14 % (p < 0.05), with fibre intake at 11.8 against 6.1 g per day and energy intake unchanged. The sample is very small and the population very specific; the result suggests a direction, it does not establish a general recommendation.

Safety and regulatory status

Agar-agar carries the number E 406. EFSA re-evaluated it in 2016 and concluded that there was no need for a numerical acceptable daily intake and no safety concern for the general population at the reported uses. In detail: agar is unlikely to be absorbed unchanged and is only slightly fermented by the gut microbiota; no concern was identified regarding genotoxicity; no carcinogenic effects were reported at the highest doses tested, 4,500 mg per kilo of body weight per day in mice and 2,500 in rats; and an oral intake of 4,500 mg per person per day, about 64 mg per kilo, was tolerated in humans for twelve weeks without noticeable side effects. The highest estimated dietary exposure reaches 26 mg per kilo per day in toddlers at the 95th percentile.

In Switzerland, the additives ordinance places agar-agar in Group I, additives authorised in line with good manufacturing practice, meaning without a general numerical limit.

There is, however, one precise prohibition, and it is instructive. Agar-agar, along with the alginates, carrageenan, locust bean gum, guar gum and several other gelling agents, may not be used in jelly mini-cup products, defined as firm-textured jelly confectionery contained in semi-rigid mini-cups or mini-capsules, intended to be swallowed in a single mouthful by squeezing the container to project the contents into the mouth.

The reason fits in one sentence: a firm gel that does not melt at body temperature, swallowed whole, does not dissolve in the airway. The property this page describes as an asset becomes a hazard in that specific format. The lesson for home cooking is not to avoid agar but to think about the format served to small children: firm bite-sized cubes are best avoided, whereas a spoonable jelly or one cut into small pieces poses no such problem.

Rescuing a failed preparation

An agar gel is reversible, which forgives a great deal.

If the gel is too soft, return it to the pan, melt it above 60 degrees, add 1 g of agar per litre pre-dispersed in a little cold liquid, bring briefly to the boil and re-mould. The important word is briefly: re-boiling an already cooked preparation at length degrades it, especially if it is acidic.

If the gel is too firm, same operation with extra water, juice or plant milk. Going from 4 to 3 g per litre means adding a third more liquid, which also dilutes the flavour: you will usually need to re-sweeten or re-season.

If the gel is grainy, with small firm specks in a liquid mass, the powder was not dispersed cold. Melt it down, blend with a stick blender, boil again for a minute.

If water has pooled at the bottom of the mould, that is syneresis rather than a dosing fault. It comes from the progressive aggregation of the double helices, which contracts the polymer network and reduces the interstitial space available to hold water. It worsens over time and with a high dose. A gel meant to keep for two or three days is better at around 3 g per litre than at 5.

Storage

The finished gel keeps two to three days in the fridge, covered so it does not dry on top or pick up odours. Beyond that, syneresis becomes visible and the texture degrades without the preparation having become unsafe: this is a quality criterion, not a safety one, since safety depends on the other ingredients in the recipe.

Freezing is not advisable. Ice crystals tear the gel network, which releases its water on thawing and collapses.

The dry powder, by contrast, keeps for a very long time in an airtight jar away from light and moisture. It does not go rancid and does not measurably lose gelling power over several years, provided it stays dry. A powder that has taken up moisture clumps and disperses badly.

Three base preparations

Fruit jelly

For 500 ml of apple, grape or pear juice, on the clear side, use 2 g of agar for a soft result or 2.5 g for a jelly that unmoulds. Disperse the agar in the cold juice while whisking, bring to the boil while stirring, hold for a minute, pour immediately into moulds. Let it set at room temperature, then refrigerate.

With an acidic juice, orange, red fruit or pineapple, it is better to boil the agar in 300 ml of water or neutral juice and add the 200 ml of acidic juice off the heat, raising the dose to 3 g.

Plant-based chocolate flan

For 500 ml of plant milk, 1.5 g of agar gives a spoonable cream and 2 g a flan that unmoulds. Mix the agar into the cold milk with 20 g of unsweetened cocoa and your chosen sweetener, whisk hard to avoid cocoa lumps, bring to the boil, hold for a minute while whisking. Pour, let it cool slightly, refrigerate. Cocoa already thickens the mixture: past 2.5 g of agar the texture turns distinctly brittle. The page on sweet silken tofu covers the alternative for a creamier result, and the one on vegan chocolate goes into choosing cocoa and how it behaves when melted.

Legume terrine

For 500 ml of blended base, cooked lentils or chickpeas, aromatics, a little stock, 2 to 2.5 g of agar is enough for a terrine that slices. Since the base is thick, whisk continuously and extend the boil to two minutes so the heat reaches everything. Pour into an oiled mould, cool, then refrigerate for at least four hours. The proportions of the base itself are covered in the guide to batch cooking legumes.

Common mistakes

Dosing by the spoon. The gap between a level and a heaped teaspoon is over 50 %. This is the leading cause of inconsistent results from one batch to the next with the same recipe.

Tipping the powder into hot liquid. The lumps that form will not redissolve and part of the agar is lost to the gel.

Extending the boil on an acidic preparation. The reflex makes worse the very problem it is meant to fix.

Judging the set too early. An agar gel starts setting at 43 degrees but takes one to two hours to reach its final firmness. Adding more agar because a preparation looks soft after twenty minutes reliably produces an over-firm gel.

Expecting a gelatin texture. No dose will give the melting quality of a gelatin jelly. When that melting quality is the point, the answer is fat or a rich purée, not an adjustment to the gelling agent.

By profile

Beginner. Start with an apple juice jelly at 2.5 g per litre, no acidity and no thickness. The result is predictable and lets you calibrate your own powder before tackling a recipe that matters.

Anyone eating vegan. Agar solves the gelling agent question, not the creaminess question. The most convincing preparations almost always pair agar with a fatty element, and aquafaba desserts offer another route to the aerated textures agar cannot produce.

Anyone cooking ahead. This is the profile agar suits best, since the gel does not collapse at room temperature. Stay around 3 g per litre to limit syneresis over two or three days.

Parents of small children. Avoid firm bite-sized cubes, for the reason set out above. A spoonable jelly or small irregular pieces are preferable.

Anyone trying to increase fibre. The few grams used for gelling will not contribute measurably. Useful fibre sources are elsewhere, notably among the pantry legumes.

Limits of this page

Gel strength of retail powders is not published by consumer manufacturers, and the doses in the table are given for an average-strength powder in a neutral liquid. A one gram per litre difference between two brands is normal.

The 32 to 43 degree setting range and the above-60 degree melting point are those reported in the literature for food-grade agar; they shift with the degree of sulfation, and therefore with the seaweed species and the extraction process.

The two clinical trials cited involve small numbers, 76 patients in one and 20 residents in the other, in specific populations. They indicate orders of magnitude, not rules that transfer to everyone.

Finally, acid hydrolysis of agarose is a well-established mechanism in polysaccharide chemistry, but there is no published table giving the loss of gelling power as a function of pH and boiling time under home kitchen conditions. Increasing the dose by a third in an acidic medium is an empirical rule of thumb, not a measured value.

Frequently asked questions

Can a dish gelled with agar be reheated? Yes, provided you go above 60 degrees to melt the gel, then let it cool below 43 degrees to set again. The gel tolerates two or three cycles, after which the texture degrades, and the more acidic the preparation the faster that happens.

Does agar-agar have a taste? It is practically neutral. A faint iodine note is sometimes noticeable in a very lightly flavoured preparation, a sweetened water jelly for instance, and disappears in any fruit or chocolate recipe.

Can it be used in a raw preparation? Not directly, since dissolving it requires boiling. The method is to boil the agar in a small fraction of the liquid, about a fifth, then whisk it quickly into the rest. The mixing has to be fast: below 43 degrees, setting begins.

How does it differ from pectin? Pectin gels in the presence of sugar and acid, which suits it to classic jams, whereas agar gels simply by cooling after boiling, independently of sugar. For low-sugar preparations agar is simpler; for a traditional jam, pectin gives a texture closer to expectation. The page on jellies and jams without gelatin compares the two in detail.

How much agar replaces one gelatin leaf? There is no universal answer, and treating conversion tables with suspicion is the best advice available. As an order of magnitude, six leaves per litre corresponds to 3 or 4 g of agar, but only for a firm moulded gel. For a mousse or a bavarois, the substitution does not work as such.

Does agar-agar help with weight loss? The only randomised trial available shows a 1.5 kg difference over twelve weeks, attributed by its authors to reduced calorie intake, with agar eaten as a bulking food. The few grams that gel a dessert have no relationship to that protocol.

Is it suitable for children? At culinary doses, yes, since EFSA identified no safety concern. The reservation concerns format: firm single-mouthful jellies are best avoided for small children.

Sources

• FAO/WHO Joint Expert Committee on Food Additives, AGAR monograph, 49th JECFA (1997), FNP 52 Add 5
• EFSA, Re-evaluation of agar (E 406) as a food additive, EFSA Journal, 2016
• Swiss Federal Food Safety and Veterinary Office, annexes to the FDHA Ordinance on additives (OAdd, SR 817.022.31)
• Menaka et al., Mini-review: red seaweed Hydropuntia edulis and agar derivatives, PeerJ, 2026
• From Sea to Sweet: Seaweed’s Role in Nutritious and Sustainable Confectionery, Comprehensive Reviews in Food Science and Food Safety, 2026
• Maeda et al., Effects of agar (kanten) diet on obese patients with impaired glucose tolerance and type 2 diabetes, Diabetes, Obesity and Metabolism, 2005
• Fiber-rich snacks improved constipation in Vietnamese elderly, Asia Pacific Journal of Clinical Nutrition, 2026