On the public record, research peptides are very pure and usually overfilled. Across 2,087 de-duplicated lab results from Kovera, Janoshik, Particle, Swiss Chems, Ascension and Injectify, median HPLC purity was 99.86% and only 7 of 1,996 results measured under 95%. Fill varies more: the median vial held 109.6% of its label, 47.6% held more than 110%, and 1.8% held under 90%. Of the 35 vials under 90% of label, 23 still measured 99% pure or better. These are results vendors chose to publish, not a random sample of the market.
Short version. Across 2,087 public peptide lab results from six sources, the median HPLC purity was 99.86%. The median vial held 109.6% of the milligrams on its label, and 47.6% held more than 110%. Only 1.8% came in under 90% of label. On the published record, a research peptide is rarely impure. When something is wrong, it is usually the amount in the vial or the identity of what is in it.
The bias you have to read first
These are results vendors and labs chose to publish. A vendor that gets a bad certificate can leave it off its website, and a buyer who never tests never shows up at all. Kovera and Janoshik both publish failures next to passes, which is why they make up most of this pool. The client still chooses which samples reach them. Every rate below is a rate among published results. The share of bad vials on the market is almost certainly higher.
Selection also explains why our numbers sit so far from the "43% failed" figure quoted around the industry. Our Janoshik 7,164-test analysis explains where that number comes from. The two figures measure different populations.
What we pooled
| Source | Results kept | With a fill figure | Median fill | Median purity |
|---|---|---|---|---|
| Kovera Labs certificates | 1,648 | 1,605 | 110.0% | 99.88% |
| Janoshik public tests | 242 | 182 | 109.3% | 99.70% |
| Particle Peptides archive | 68 | 68 | 98.5% | 99.34% |
| Swiss Chems test page | 44 | 21 | 101.0% | 99.55% |
| Ascension Peptides COA page | 64 | 26 | 112.2% | 99.75% |
| Injectify test reports | 21 | 19 | 105.5% | 99.67% |
| All sources | 2,087 | 1,921 | 109.6% | 99.86% |
Kovera and Janoshik are lab-run public databases. The other four are vendor-published certificate pages, tested by outside labs. Test dates run from July 2023 to July 2026, but 1,831 of the 1,921 fill results are dated 2026. The pool covers 187 distinct client names. That count runs a little high, because some vendors appear under both a brand and an email address.
Per-lab detail is on our Kovera Labs analysis and the Janoshik page above. This page pools them to answer one question: how pure, and how full, are research peptides on the public record?
What we kept, and how we removed duplicates
We started from 2,350 mirrored certificates and left out 133 that are not peptide potency results: anabolic steroids, SARMs, nootropics and other small-molecule drugs, bacteriostatic and sterile water, safety-only panels (sterility, endotoxin, heavy metals) and listings with no readable result. We kept peptides, peptide blends, protein hormones such as hGH, and the three non-peptide compounds peptide shops sell next to them: NAD+, glutathione and 5-Amino-1MQ.
Labs often issue one test result several times, once for each client that pays to have its name on the certificate. If we counted every copy, one sample would count as six. We treated two results as the same test when the source, compound, measured milligrams, purity and test date all matched, and kept one. We also dropped re-listed copies of the same order in Particle's archive. That removed 130 copies, leaving 2,087.
Two examples from the record. Janoshik tests 137807 and 147736 to 147740 are six test numbers issued to six different clients, all for a 100 mg GHK-Cu vial measuring 77.27 mg at 99.435% purity, dated 14 April 2026. We count that as one result. In Kovera's records, one retatrutide result (33.02 mg, 99.966%, 28 February 2026) appears on six certificates, KVR-2026-88383C and five reissues, under three client names. We count that as one result too.
Two more exclusions from the fill figures. One certificate listed a 1 mg label against 1,024 mg measured, which is a unit error. Four Kovera certificates found 0 mg of the labelled compound. Those are identity failures, counted as such, and left out of the fill median.
Purity and fill are two different numbers
Purity is an HPLC area percentage. The lab runs the sample through a column, records a UV trace, and reports what share of the peak area belongs to the target peptide. 99.86% purity means 99.86% of the signal the detector saw was the right molecule. It does not say how much of it is in the vial.
Content, or fill, is a separate quantification. The lab measures how many milligrams of the peptide are present, usually against a reference standard, and the certificate prints that next to the label claim. We report it as a percentage of label: 10.4 mg in a 10 mg vial is 104%.
So a vial can be 99.9% pure and half empty. In our pool, 35 results came in under 90% of label. 23 of those 35 measured 99% pure or better. A Kovera certificate for client Zyron Research (KVR-2026-B9B935, 27 February 2026) prints 412.84 mg of NAD+ in a 500 mg vial, or 82.6%, at 99.716% purity. One for client RedlinePeppers (KVR-2026-7449A0, 21 May 2026) prints 11.44 mg of GHK-Cu in a 50 mg vial, or 22.9%, at 99.537%.
The reverse also happens, though rarely. A Kovera certificate for client Axon Peptide (KVR-2026-E91AD0, 20 May 2026) shows a full vial of BPC-157, 5.64 mg against a 5 mg label, at 46.849% purity. Kovera graded it a fail.
Client names on this page are as printed on the lab's public certificate, read on the date of our mirror. The client is whoever commissioned the test, which is not always the seller of a given vial, and a certificate can carry a reporting error. Each report number can be re-checked in our Kovera index.
If you only read the purity line on a certificate, you will miss most of what goes wrong. Our COA verifier checks the measured mass against the label for you.
How pure are research peptides?
Across the 1,996 results with a purity figure:
| Purity band | Results | Share |
|---|---|---|
| 99% or higher | 1,865 | 93.4% |
| Under 99% | 131 | 6.6% |
| Under 98% | 30 | 1.5% |
| Under 95% | 7 | 0.4% |
The bottom tenth of results starts at 99.2%. Of the seven results under 95%, three are well below: the Axon Peptide BPC-157 certificate above, a Peptide Partners epitalon certificate from Kovera at 72.846% (20 March 2026), and a Janoshik sermorelin test at 80.583% (14 May 2026). Two of the seven are human growth hormone, a protein rather than a short peptide.
Purity does not cover contamination either. Endotoxin, sterility and heavy metals are separate panels. Our endotoxin guide covers how to read them when they are present.
How full are peptide vials?
Across the 1,921 results with a label and a measured mass:
| Fill band | Results | Share |
|---|---|---|
| Under 90% of label | 35 | 1.8% |
| Under 95% of label | 93 | 4.8% |
| Under 100% of label | 224 | 11.7% |
| 90% to 110% of label | 972 | 50.6% |
| Over 110% of label | 914 | 47.6% |
| Over 120% of label | 98 | 5.1% |
The middle half of results sits between 105.9% and 114.4% of label. Overfill is the norm: you are far more likely to see 11 mg on a 10 mg certificate than 9 mg.
No vendor in the pool explains its overfill, so what follows is our inference. Lyophilised peptide is weighed and dispensed in small amounts, some is lost to the vial wall and the filling equipment, and the assay itself carries a few percent of uncertainty. A vendor aiming at exactly 100% would see a share of its certificates land under label. Aiming 10% high makes that much less likely. The data fits that explanation, but no vendor statement confirms it.
The vendor-published sets look different from the lab databases. Particle Peptides' own archive has a median fill of 98.5% and holds 18 of the 35 sub-90% results, 12 of them dated 2023 and 2024. Particle publishes its full archive, old lots included, which is part of why those results are visible at all.
Purity and fill by compound
Every compound or named blend with 20 or more results. "Under 90%", "under 95%" and "over 110%" are shares of results with a fill figure.
| Compound | Results | Median fill | Under 90% | Under 95% | Over 110% | Median purity |
|---|---|---|---|---|---|---|
| Retatrutide | 323 | 110.2% | 0.7% | 2.9% | 51.8% | 99.86% |
| Tirzepatide | 187 | 111.1% | 0.0% | 0.6% | 56.3% | 99.88% |
| GHK-Cu | 104 | 108.3% | 2.0% | 3.1% | 31.6% | 99.88% |
| Tesamorelin | 96 | 109.1% | 1.1% | 3.4% | 46.1% | 99.89% |
| MOTS-c | 92 | 109.6% | 1.1% | 4.4% | 46.7% | 99.90% |
| BPC-157 | 90 | 111.1% | 1.3% | 6.3% | 56.3% | 99.88% |
| BPC-157 + TB-500 | 79 | 111.3% | 1.5% | 2.9% | 58.8% | 99.77% |
| NAD+ | 74 | 104.4% | 5.6% | 11.3% | 11.3% | 99.85% |
| CJC-1295 + ipamorelin | 68 | 111.5% | 0.0% | 0.0% | 68.3% | 99.79% |
| KLOW blend | 58 | 108.7% | 3.9% | 3.9% | 25.5% | 99.80% |
| Semax | 44 | 109.1% | 2.4% | 4.8% | 45.2% | 99.93% |
| Ipamorelin | 43 | 109.7% | 2.7% | 16.2% | 48.6% | 99.83% |
| GLOW blend | 41 | 107.9% | 2.7% | 2.7% | 27.0% | 99.79% |
| Selank | 41 | 112.6% | 0.0% | 5.3% | 60.5% | 99.92% |
| Semaglutide | 41 | 111.5% | 0.0% | 0.0% | 57.9% | 99.92% |
| KPV | 37 | 111.3% | 2.9% | 8.8% | 50.0% | 99.92% |
| Melanotan-2 | 37 | 108.1% | 9.1% | 18.2% | 42.4% | 99.87% |
| SS-31 | 37 | 108.5% | 0.0% | 2.9% | 45.7% | 99.88% |
| PT-141 | 34 | 109.9% | 6.3% | 9.4% | 46.9% | 99.88% |
| TB-500 | 34 | 111.4% | 0.0% | 3.1% | 53.1% | 99.89% |
| DSIP | 32 | 108.9% | 3.4% | 6.9% | 44.8% | 99.86% |
| Thymosin alpha-1 | 32 | 113.2% | 0.0% | 3.3% | 66.7% | 99.70% |
| Epitalon | 31 | 109.0% | 3.3% | 6.7% | 40.0% | 99.70% |
| CJC-1295 no DAC | 30 | 109.6% | 3.7% | 3.7% | 48.1% | 99.77% |
| Cagrilintide | 30 | 111.4% | 0.0% | 3.4% | 55.2% | 99.91% |
| 5-Amino-1MQ | 29 | 113.1% | 0.0% | 3.7% | 63.0% | 99.88% |
| Glutathione | 29 | 98.7% | 0.0% | 22.2% | 3.7% | 99.92% |
| Kisspeptin | 25 | 110.9% | 0.0% | 4.5% | 59.1% | 99.90% |
| Sermorelin | 25 | 111.8% | 0.0% | 0.0% | 59.1% | 99.92% |
| AOD-9604 | 22 | 110.2% | 9.5% | 14.3% | 52.4% | 99.84% |
We grouped code names with their parent compound, so a vial sold as "GLP-3 RT" counts as retatrutide. Our retatrutide lab-record analysis uses the same 323 results. Small groups move a lot on one result. AOD-9604's 9.5% under 90% is two results out of 21.
Glutathione: the one compound under label
Glutathione is the only compound in the table whose median vial holds less than the label says: 98.7% across the 27 results with a fill figure. Six of the 27 came in under 95%, and only one of 27 was over 110%, against 47.6% across the whole pool. All six were Kovera certificates from April and May 2026, with these clients printed: Ion Peptide (90.8%), Axon Peptide (91.8%), BioEdge Research Labs (92.8%), Amino 88 (93.1%), Instant Peptides (94.8%) and one certificate with no client name (94.4%). Purity on all 29 glutathione results was high, with a median of 99.92% and a low of 98.60%.
Glutathione and NAD+ are mostly sold in 500 to 1,500 mg vials, many times the mass of a typical 5 to 10 mg peptide vial. Our inference: a 10% overfill on a 1,500 mg vial costs a vendor far more than on a 10 mg vial, so the overfill buffer shrinks. Kovera graded all six short glutathione results a pass.
NAD+: lowest fill among the high-volume compounds
NAD+ has a median fill of 104.4% across 71 results with a fill figure. Eight came in under 95% and four under 90%. Only 11.3% were over 110%. The lowest is a Kovera certificate for client RedlinePeppers (KVR-2026-82AC32, 21 May 2026) printing 10.94 mg in a 500 mg vial, 2.2% of label, at 98.208% purity and graded a pass. A figure that low may be a reporting error on the certificate. We cannot tell from the record, and we report the certificate as we retrieved it on 2 August 2026. Ascension Peptides' own archive shows a 1,000 mg NAD+ vial at 784.32 mg (78.4%) on 26 January 2026. One NAD+ certificate, client Instant Peptides on 11 May 2026, did not confirm identity.
GHK-Cu
GHK-Cu sits at 108.3% median fill and 99.88% median purity across 104 results, with 31.6% over 110%, the lowest overfill rate among the compounds with 100 or more results. Its two sub-90% results are the RedlinePeppers certificate above and the six-client Janoshik result counted once. Our GHK-Cu guide counts 102, because it also folds reissues whose test dates differ by a few days; the medians are the same.
Fails and identity failures
Across the 2,087 results, 19 were graded an explicit fail and 11 did not confirm the identity of the labelled compound. All 11 are Kovera certificates from May 2026. Four found 0 mg of the labelled compound, among them two certificates from one client dated 22 May 2026, for retatrutide and tesamorelin. Others reported over 99.8% purity for a compound that did not match the label, including "GLP-3 RT" for client Instant Peptides and "GLA-3 RT" for client Glacier Aminos, both dated 11 May 2026.
Purity cannot catch an identity failure, because a wrong molecule can still produce one clean peak. Labs usually confirm identity by mass spectrometry. A certificate that reports purity without an identity result has skipped that check. Our lab comparison lists which labs run which methods.
How to use this when you read a certificate
- Read the measured milligrams, not only the purity. On the published record, purity under 98% is rare (1.5%). A short fill is about as rare, but 28 of the 35 sub-90% results were still graded a pass.
- Expect about 105% to 115% of label. That is the middle half of 1,921 results. A vial at 95% to 100% is within label but below the norm. Under 90% is in the bottom 2%.
- Check identity separately. 11 results in this pool failed it.
- Verify the certificate at the lab. Copies on vendor sites can be edited or reused. Our guide to verifying a Janoshik COA walks through the lab's own lookup.
- Match the lot. A certificate covers one tested sample. It tells you about your vial only if the lot number matches.
What we read, and what we did not
We read 2,350 published certificates from six sources: Kovera Labs' public verifier (mirror of 2 August 2026), Janoshik's public test listing (30 June 2026), and the certificate pages of Particle Peptides, Swiss Chems, Ascension Peptides and Injectify (May 2026). We left out 133 non-peptide or safety-only records and removed 130 duplicate copies, leaving 2,087 results. We did not include Finnrick, and we did not test any vial. We do not publish client email addresses that appear on some certificates.
The per-compound table is a CSV file on our data page, free to reuse with credit.
Frequently asked
How pure are research peptides?
In the public lab results we pooled, very pure. The median HPLC purity across 1,996 results with a purity figure was 99.86%. 131 (6.6%) were under 99%, 30 (1.5%) under 98% and 7 (0.4%) under 95%. Vendors choose which results to publish, so the unpublished population is likely worse.
What is the difference between peptide purity and peptide content?
Purity is the share of the HPLC signal that belongs to the target peptide. Content, or fill, is how many milligrams of it the lab found in the vial, compared with the label. A vial can be 99.9% pure and hold half its labelled amount. In our pool, 23 of the 35 vials under 90% of label measured 99% pure or better.
How accurate are peptide vial fills?
Usually generous. Across 1,921 results with both a label and a measured mass, the median vial held 109.6% of label and the middle half sat between 105.9% and 114.4%. 35 results (1.8%) came in under 90% and 98 (5.1%) over 120%.
Why are so many peptide vials overfilled?
No vendor in our pool states a reason. Our inference is that vendors overfill so a vial still measures at or above label after weighing error, losses in filling and normal assay variation. 914 of 1,921 results (47.6%) were more than 10% over label.
Which peptides are most often underfilled?
Glutathione is the only compound with 20 or more results whose median fill sits below label: 98.7% across 27 results with a fill figure, with 6 under 95%. NAD+ is next at 104.4% across 71, with 8 under 95% and 4 under 90%. Both are mostly sold in 500 to 1,500 mg vials, far larger than most peptides.
Does a passing peptide COA mean the vial is full?
No. 28 of the 35 results under 90% of label in our pool were still graded a pass, including a 500 mg NAD+ certificate that printed 10.94 mg. Read the measured milligrams on the certificate, not only the purity line and the pass mark.