Supplement Forms Compared: Creatine, Magnesium and Collagen by Weight and by Evidence

Formula weight decides how much of each compound is the part being paid for. Creatine monohydrate is 87.9% creatine and creatine hydrochloride 78.2%; magnesium oxide is 60.3% magnesium, citrate 16.2%, bisglycinate 14.1% and L-threonate 8.3% — ceilings by arithmetic, not laboratory measurements. On evidence, monohydrate is the form 275 of 343 randomised creatine trials used, and no other form has beaten it head to head.

Three tables follow. The creatine and magnesium tables are computed the same way: the weight of creatine or of magnesium in one formula unit, divided by the formula weight PubChem lists for the whole compound. Collagen gets a different kind of table, because collagen sources differ by tissue and type rather than by a salt partner. Every row carries what the research found beside the arithmetic, and the last section links every comparison on this site, so this page is the index as well as the reference.

The formulas and weights are PubChem’s, retrieved on 17 October 2026, and the PubChem compound number is printed in every row so that any figure can be checked in a minute. The creatine column is not ours alone: a 2022 review computed the same fractions from the same database — monohydrate 87.9%, hydrochloride 78.2%, nitrate 67.5%, pyruvate 59.8% — in its Table 1 (Kreider et al., Nutrients 2022;14(5):1035).

Which form of creatine is best?

Creatine monohydrate, on both counts that can be checked. It carries 87.9% creatine by formula weight, more than any form except anhydrous creatine, which is the same molecule with the water removed. It is the form 275 of 343 randomised creatine trials used, in a 2026 audit, and the ISSN’s 2017 position stand says clinical evidence has not demonstrated that citrate, ethyl ester, buffered or nitrate forms promote greater creatine retention than monohydrate.

Two rows in that table carry more weight than the rest. The ethyl ester was sold on better uptake, and the 2009 head-to-head — Spillane and colleagues, J Int Soc Sports Nutr 2009;6:6, 30 men new to resistance training over seven weeks — found serum creatine lower than with monohydrate and serum creatinine higher, which is the breakdown product rather than the compound being sold. The authors concluded the ester was “not as effective at increasing serum and muscle creatine levels”. Their own muscle measurements showed no significant difference between the two creatine groups, so the conclusion runs slightly ahead of its data, and it is printed here with that attached.

The hydrochloride was sold on solubility, and the solubility is real: it dissolves far more readily in water. The figure attached to it, 38 times, came from a laboratory comparison that measured monohydrate at a pH of 8.6 and hydrochloride at a pH of 0.3, according to the 2022 review, which adds that dissolving creatine in acid “would have no effect on bioavailability”. When the review was written it found no PubMed-indexed study of the hydrochloride and muscle creatine or performance, and it called claims that the salt is more bioavailable, effective or safe than monohydrate “not supported”. The head-to-head trials published since are weighed on the hydrochloride page.

Read the funding with the findings, because it runs in both directions. The buffered-creatine trial — Jagim and colleagues, J Int Soc Sports Nutr 2012;9(1):43 — was paid for by a creatine monohydrate manufacturer and found no advantage for the buffered form. The 2007 plasma study of citrate and pyruvate — Jager and colleagues, J Int Soc Sports Nutr 2007;4:17 — was funded by a creatine manufacturer that supplied all three powders, and its abstract describes the pyruvate peak as significantly higher than monohydrate’s while the paper’s own table gives p = 0.075. The nitrate trial — Galvan and colleagues, J Int Soc Sports Nutr 2016;13:12 — was paid for by a supplement company, with one of its research directors among the authors, and found nitrate similar rather than better. The 2022 review that graded every marketed form placed monohydrate alone in its strong-evidence tier, put the ester, hydrochloride, nitrate, citrate, pyruvate, buffered and chelated forms in a middle tier, and found none of them better than monohydrate; its publication charge was funded by a creatine monohydrate manufacturer, and one of its authors is acting chair of that company’s scientific advisory board on creatine. The arithmetic column needs nobody’s trust, which is why it comes first.

Which creatine form carries the most creatine per gram?

Anhydrous creatine, at 100%, because nothing is attached to it. Monohydrate follows at 87.9%, the ethyl ester at 82.4% and the hydrochloride at 78.2%, and every salt pays for its partner: nitrate is 67.5% creatine, tricreatine citrate 67.2% and pyruvate 59.8%. Matching the 4.4 g of creatine in 5 g of monohydrate takes 5.6 g of hydrochloride, 6.5 g of nitrate or 7.3 g of pyruvate.

Two cautions keep that ranking honest. More creatine per gram is a statement about volume, not about effect, and the 2022 review’s note on the anhydrous form is that heat-drying leaves more creatinine behind, so the most concentrated form is not automatically the cleanest. And the names on labels are looser than the chemistry. A bottle reading “creatine citrate” can hold the 1:1 salt at 40.6% or tricreatine citrate at 67.2%, and the ethyl ester also exists as a hydrochloride salt, PubChem 25114309, at 67.0% creatine. If the label does not say which, the division cannot be run. The hydrochloride division, the ester’s record and the buffered trial each have a page of their own.

What percentage of each magnesium form is elemental magnesium?

By formula weight: oxide 60.3%, carbonate 28.8%, citrate 16.2%, malate 15.5%, bisglycinate 14.1%, lactate 12.0%, hydrated chloride 12.0% and L-threonate 8.3%. A Supplement Facts panel declares the element, and the compound weight is the larger number, so dividing one by the other shows which forms a capsule could be. A fraction above a form’s ceiling means the capsule is not what its label says.

The column the NIH Office of Dietary Supplements fills is narrower than it is usually quoted. Its sentence reads: “Small studies have found that magnesium in the aspartate, citrate, lactate, and chloride forms is absorbed more completely and is more bioavailable than magnesium oxide and magnesium sulfate.” That names four forms on one side and two on the other, and glycinate, malate, carbonate and threonate on neither. It supports a direction — forms that dissolve well tend to be absorbed more completely — and it does not supply a ranking of the forms in this table. Nothing here should be read as one.

Why do two labels for the same magnesium form show different percentages?

Because the compound in the capsule is rarely the pure, dry salt in the table. Water of crystallisation lowers the fraction — magnesium chloride is 25.5% magnesium when anhydrous and 12.0% as the hexahydrate — and so does a different salt ratio, a carrier or an undisclosed blend. A figure under a form’s ceiling is ordinary. A figure above it is the one to question.

Citrate shows the ratio problem in a single word. Trimagnesium dicitrate, the anhydrous salt in the table, is 16.2% magnesium; magnesium hydrogen citrate, PubChem 8949, is 11.3%; both answer to the name magnesium citrate. Lactate drops from 12.0% to 10.2% as the dihydrate. Run on our own panel, the division lands where it should: Night Work declares 275 mg of elemental magnesium from 2,500 mg of magnesium glycinate, which is 11.0% — under the 14.1% a pure, dry bisglycinate would reach, not over it. The same division run against oxide, against citrate and against chloride are on their own pages.

Which collagen source is best: bovine, porcine, marine or chicken?

None has been shown to beat another, because the trials this site cites never compared animals. The real differences are type, allergen and processing: hide and bone carry mainly type I, hide with type III as well, fish yields type I, cartilage carries type II, fish is a major food allergen that a US label must name, and hydrolysis turns any of them into peptides that dissolve rather than gel.

Collagen is a family rather than a molecule. A 2019 review of collagen sources in Molecules places type I in bone, tendon and ligament, type II in cartilage and type III in muscle and blood vessels, and puts hydrolysed collagen at roughly 3 to 6 kDa — short peptides. The same review names the collagen type and the source it is extracted from as the main factors setting those peptides’ size and solubility, and notes religious considerations among the limits of bovine and porcine sources.

Type II from chicken breastbone cartilage is sold two ways, and they are different products. Hydrolysed type II is a collagen peptide powder like the others in the table. Undenatured type II is a different product, sold at milligram doses with a literature of its own, and this site makes no statement about it in either direction. What a label must say also differs by animal: fish is one of nine major food allergens, and US labelling law requires the species to be named, according to the FDA. The source comparison is taken further on bovine against marine collagen and what the type numbers mean.

Does the form change how much you need to take?

Only by the arithmetic of its partner, unless a trial shows a form working at a smaller dose of the active, and for creatine none has. The creatine in 5 g of monohydrate takes 5.6 g of hydrochloride or 7.3 g of pyruvate to match; 275 mg of magnesium takes 456 mg of oxide, 1,701 mg of citrate or 1,951 mg of bisglycinate. A smaller serving of a lighter form is a smaller dose.

The claim worth checking is the lower-dose one, because it is the one doing work on the price. The 2012 buffered-creatine trial ran it directly: 1.5 g a day of the buffered product, the maker’s recommended dose, against loading and maintenance doses of monohydrate. In the 25 men with usable biopsies, unbound muscle creatine rose 22.3 mmol per kg of dry muscle by day 28 on monohydrate and 4.7 on the buffered product at its label dose, a gap that did not reach significance (group by time, p = 0.46) with eight or nine men in each arm. Weak evidence, all of it pointing one way. The 2022 review puts the principle plainly: the creatine in any salt “must be equalized” to monohydrate to deliver the same amount.

Are these percentages measured by a laboratory?

No. Every percentage on this page is arithmetic: the weight of creatine or magnesium in one formula unit divided by the formula weight PubChem lists for the compound. It is the ceiling a pure, dry compound can reach. What a particular lot of powder actually contains is a laboratory figure, printed on a certificate beside the specification it was measured against, and it is a different kind of number.

The two kinds of number do different jobs. Formula weight says what a compound can be; an assay says what a sample was. Base Work’s raw-material certificate, from an independent laboratory, reports creatine monohydrate at 99.87% against a greater-than-99.5% specification, by HPLC — a measurement of one raw-material lot, carried out for another company, not of a finished TARE tub, and not a formula-weight ceiling either. How to read that kind of document is on what a certificate of analysis is, and every row of ours is transcribed on Testing & Certification.

What the research does not show

No trial shows a creatine salt raising muscle creatine more than monohydrate does at an equal dose of creatine, and that is the most this page can say for monohydrate rather than the least. The head-to-head record is short and small: one dose in six adults for citrate and pyruvate, seven weeks in 30 men for the ester, 28 days in 36 participants for the buffered form and 28 days in 48 for nitrate. The ester and buffered trials enrolled young adults, with mean ages near 20, and none of the four ran longer than seven weeks.

The audit behind the 80.2% figure found the comparison itself to be rare. Of 343 randomised creatine trials published from 1994 to 2023, five compared more than one formulation, and the authors judged those comparisons “methodologically limited, particularly because of non-equimolar dosing” — a different amount of creatine in each arm. The same audit found that 15.5% of trial reports never said which form of creatine was used (Nutrients 2026;18(16):2697).

On magnesium, the absorption evidence the Office of Dietary Supplements summarises is, in its own words, from small studies, and glycinate, malate and threonate are not in it. On collagen, no trial this site cites compared one animal source with another for any outcome, and the evidence on collagen and connective tissue, with every limit attached, is on the collagen evidence table. And a formula weight says nothing about absorption, tolerance or effect. It is the first number to check, not the last; how to read the studies that come after it has a page of its own.

Where our three products sit in these tables. Base Work is creatine monohydrate, one ingredient, 5 g a day — the reference row of the first table. Night Work declares 275 mg of elemental magnesium from 2,500 mg of magnesium glycinate per three-capsule serving, 11.0% by the label’s own figures. Under Load is collagen peptides, type I and III, from grass-fed bovine sources in Brazil — a description of the source, not a claim that the animal makes the difference. What each certificate does and does not cover is on Testing & Certification.

Every comparison on this site

Each page below takes one comparison further than a table can. New pages and long-standing ones are listed together, by ingredient.

Creatine

• Creatine research summary — the key trials and position stands in one table, each with its limits.
• Creatine HCl vs monohydrate — 78% against 88%, and what the solubility argument does and does not show.
• Micronized creatine vs monohydrate — the same molecule, milled to a finer particle.
• Creatine ethyl ester vs monohydrate — sold on uptake, and what the head-to-head measured.
• Buffered creatine vs monohydrate — the lower-dose claim, run against monohydrate.
• Monohydrate against a trademarked grade — what an ingredient trademark does and does not tell a buyer.
• Do creatine gummies work? — the format, its stability problem and what the panel has to show.
• Creatine vs whey protein — two different jobs, not rivals.
• How to choose a creatine supplement — the checklist, in order.

Collagen

• Collagen research summary — the tendon, strength and muscle trials, each with its limit.
• Bovine vs marine collagen — types, allergens and what the research used.
• Collagen types explained — what type I, II and III name once the protein is hydrolysed.
• Collagen peptides vs gelatin — the same protein, broken down to a different degree.
• Collagen vs bone broth — why a recipe cannot tell you a dose.
• Bioactive collagen peptides vs collagen peptides — what a peptide-profile trademark buys.
• Collagen vs whey protein — which one is a complete protein, and which one we sell.
• How to choose a collagen supplement — protein per serving, dose and documents.

Magnesium

• Magnesium research summary — roles, intake, absorption studies and their limits.
• Magnesium glycinate vs oxide — 14% against 60%, and why the bigger number is not the better form.
• Magnesium glycinate vs citrate — two soluble forms, 14% and 16%.
• Magnesium glycinate vs malate — 15%, and the story told about malic acid.
• Magnesium chloride vs glycinate — hydrated chloride at 12%, and magnesium applied to the body.
• Magnesium bisglycinate vs glycinate — one word, one compound.
• Magnesium threonate vs glycinate — 8% against 14%.
• How much magnesium per day — the 420 mg allowance and the 350 mg supplemental ceiling.
• How to choose a magnesium supplement — the elemental number, the form and the total.
• Can you take too much magnesium? — how a stack reaches 350 mg without anyone noticing.

Labels, certificates and studies

• Supplement label glossary — the terms on a panel, a certificate and a study, defined.
• How to read supplement research — design, size, certainty and who paid.
• Supplements for men over 45 who lift — an evidence-ordered list, including what to skip.
• How to read a supplement label — the panel, line by line, in sixty seconds.

Sources

1. National Center for Biotechnology Information. PubChem Compound database: molecular formulas and weights for the compound numbers tabled (retrieved 17 October 2026). pubchem.ncbi.nlm.nih.gov
2. Kreider RB, Jager R, Purpura M. Nutrients 2022;14(5):1035. doi:10.3390/nu14051035
3. Kreider RB, Kalman DS, Antonio J, et al. J Int Soc Sports Nutr 2017;14:18. jissn.biomedcentral.com
4. Bucevic Popovic V, Sovulj A, Raic S, et al. Nutrients 2026;18(16):2697. doi:10.3390/nu18162697
5. Spillane M, Schoch R, Cooke M, et al. J Int Soc Sports Nutr 2009;6:6. PMC2649889
6. Jagim AR, Oliver JM, Sanchez A, et al. J Int Soc Sports Nutr 2012;9(1):43. jissn.biomedcentral.com
7. Galvan E, Walker DK, Simbo SY, et al. J Int Soc Sports Nutr 2016;13:12. jissn.biomedcentral.com
8. Jager R, Harris RC, Purpura M, Francaux M. J Int Soc Sports Nutr 2007;4:17. jissn.biomedcentral.com
9. NIH Office of Dietary Supplements. Magnesium: Fact Sheet for Health Professionals (updated 6 January 2026). NIH Office of Dietary Supplements
10. Leon-Lopez A, Morales-Penaloza A, Martinez-Juarez VM, et al. Molecules 2019;24(22):4031. doi:10.3390/molecules24224031
11. US Food and Drug Administration. Food Allergies (content current as of 11 March 2026). fda.gov