Which peptides fail lab tests most often: fail rates by peptide from published test data

Which peptides fail lab tests most often

In Finnrick’s published data, CJC-1295 fails most often among widely tested peptides: about 72% of classifiable samples failed. TB-500 (63%), tesamorelin (54%) and BPC-157 (50%) follow. At the other end, retatrutide (22%), tirzepatide (21%) and GHK-Cu (13%) fail least. The three most tested products are also the three with the lowest fail shares.

The data come from Finnrick, a US platform that sends community-submitted and self-purchased samples to partner labs and publishes the results. Every count below uses Finnrick’s default standard, Purity Focus: identity correct, HPLC purity at least 99.5%, and fill within ±35% of the label claim. A different standard gives different numbers, and a fail is not the same as a fake.

Peptide fail rate by product

Fail share is failed tests divided by passed plus failed. Incomplete tests are left out, as Finnrick does. The table includes every product with at least 100 published tests.

Product	Published tests	Passed	Failed	Fail share	Data to
CJC-1295	294	83	210	72%	20 Sep 2026
TB-500	282	103	175	63%	25 Sep 2026
Tesamorelin	604	274	326	54%	20 Sep 2026
BPC-157	791	391	395	50%	20 Sep 2026
Cagrilintide	194	105	88	46%	25 Sep 2026
Melanotan II	182	110	67	38%	25 Sep 2026
PT-141	151	95	53	36%	25 Sep 2026
Semaglutide	373	250	119	32%	25 Sep 2026
Ipamorelin	441	299	139	32%	25 Sep 2026
Retatrutide	3,858	2,988	824	22%	20 Sep 2026
Tirzepatide	2,478	1,930	522	21%	20 Sep 2026
GHK-Cu	807	695	104	13%	20 Sep 2026

Rows dated 20 September come from our market research snapshot of Finnrick’s public data summary (data to 20 September 2026). Rows dated 25 September come from the same summary read on 4 October 2026, when the latest published test was dated 25 September. For the six table products that appear in both, the later summary gives the same order, with fail shares within two points.

Smaller sets are too thin to rank with confidence. MOTS-c had 77 tests with 52 fails (70% of classifiable samples) in the later summary. NAD+, a coenzyme rather than a peptide, had 27 tests with 20 fails in the September snapshot. The guide to less-tested compounds covers what a lab checks for these.

Across the seven products in the September snapshot, 2,401 of 8,767 classifiable tests failed, about 27%. That fits Finnrick’s own February 2026 line that “roughly one in three fail basic expectations”.

What “fail” means under each Finnrick standard

Finnrick classifies each sample as pass, fail or incomplete against the requirements of a chosen standard: identity, purity, fill accuracy and, in some standards, batch labelling. It publishes five standards. The counts above use the first.

Standard	Access	Purity at least	Fill vs label	Labelling	Endotoxin
Purity Focus (default)	Public	99.5%	±35%	Not graded	Not graded
Finnrick Standard	Public	99.5%	±25%	Batch labelling required	Not graded
Purity Focus Strict	Members	99.8%	±35%	Not graded	Not graded
Super Strict	Members	99.8%	±15%	Strictest labelling score	Not graded
Accuracy Focus	Members	98%	±20%	Batch labelling required	Fails above 40 EU per sample, when a result exists

The labelling score rewards a batch ID printed on the vial (2 points) over one on a paper insert (1) or only on a website (0).

Three consequences follow for reading the table:

• Purity is strict, fill is loose. A vial of the right peptide at 99.4% purity fails Purity Focus. A vial 30% under its label passes it, but fails the Finnrick Standard and Super Strict.
• Overfill fails too. A vial with 40% more peptide than the label fails, the same as one with 40% less. Vendors and users dispute this rule; Finnrick keeps it.
• Endotoxin is graded in one standard only. Under Purity Focus, a sample with high endotoxin can still pass. Endotoxin needs its own test. See the guide to endotoxin testing.

What the high-fail products have in common

The published counts do not say which requirement each failed sample missed. So the data cannot tell you whether CJC-1295 fails on identity, purity or fill. They support only careful observations.

Two of the four products with the highest fail shares have a naming problem. “CJC-1295” covers two molecules, with and without DAC. “TB-500” is used for full-length thymosin beta-4 and for short fragments. A lab tests for one specific molecule. A report is only useful when it says which one, as our CJC-1295 and TB-500 pages do. This does not prove that naming causes the failures, but it is the first thing to check on any report for these products.

Retatrutide and tirzepatide, near the bottom of the table, have the largest data sets. In the later summary, retatrutide had results from 378 vendors and tirzepatide from 262, against 69 for CJC-1295, 58 for TB-500 and 43 for cagrilintide. Volume narrows the uncertainty of a fail share, but it does not explain why a fail share is low. A product sold by few vendors can also swing on one vendor’s batches, so a small set deserves less weight than a large one.

Caveats before you compare peptides

Which samples get tested

Finnrick’s samples come from anonymous public submissions and from its own purchases, which it uses as anchor samples. Buyers choose what to send, so the mix is not a random sample of the market. Since April 2026, free testing needs a batch ID with photo evidence.

Vendors can also pay for tests through Finnrick’s Launch programme. The vendor ships four samples and two are tested. Finnrick’s vendor page says the vendor can “review and approve the test results before publishing”. In a September 2026 interview, Finnrick’s chief executive said vendors can reject publication of launch results. Finnrick states that leaderboard placement cannot be bought and that it does not accept payment to suppress failures. Vendor-submitted samples are labelled as such.

Which lab ran the test

One lab, Krause Analytical, ran 6,372 of the 10,405 tests Finnrick had published by 20 September 2026, about 61%. Pass rates on recent tests at Finnrick’s partner labs range from 33% to 74%. In our reading, those lab rates mix different vendors and periods, so they do not measure lab bias. Finnrick says it estimates per-lab offsets from overlapping tests, but it does not publish them.

Labs do differ on mg. In graymarket’s comparison of one retatrutide batch (3 September 2025), Janoshik and Chromate reported about 30 mg and Krause reported 26 mg on both its vials, about 15% lower. Peptide Critic compared five labs on another retatrutide batch and found 10.18 to 12.80 mg, a 26% spread. Krysia’s Substack sent one tirzepatide batch labelled 60 mg to three labs and got 62.3 to 75.44 mg. Each is a single published case, not proof that any lab is wrong. With a ±35% fill window, an offset of 15% will not fail a vial that holds its label amount, but it can move a vial that is already low or high across the line. The guide on why labs disagree covers the causes.

Time and recency

The data run from 17 December 2024 to September 2026. A product’s fail share mixes old and new batches. Finnrick weights recent tests more heavily in its vendor ratings, with a half-life of about two years, but the raw counts above are not weighted.

No reasons, no amounts

A fail share says how often samples missed a threshold, not by how much. A 99.4% sample and an empty vial each count as one fail. For that detail, you need the individual results.

Other published failure data

Other rating sites take a different approach. In our September 2026 review, Peptigrity listed the average purity of samples per lab, Disclosed Labs graded labs and vendors, Merit Verified scored lab quality, and PeptideBenchmark tracked reused and checkable reports. None published pass and fail counts per product, so they cannot extend the table above. The guide to how Finnrick works covers its method in more depth.

How to use peptide fail rates

A fail share is a starting point, not a verdict on your vial. Products near the top of the table are where the standard test matters most:

1. Identity by LC-MS. It catches the wrong peptide or an empty vial. See identity testing.
2. Content per vial. It catches underfill and overfill against the label. See content per vial.
3. Purity by HPLC. Then read the number against the threshold you care about, not only a pass badge.

For a batch shared by many buyers, a group test with three or more vials from different kits shows how much vials vary. The guides on fake peptides and underdosed vials explain each failure type.

What HK Analytical does

For the peptides on our compounds page, we run identity (LC-MS), purity (HPLC) and content per vial as the standard test, plus endotoxin, Sterility (USP <71>, 14 days), residual solvents, counter-ion and water content in any mix. Results within days of receipt. Sterility (USP <71>, 14 days) is the one test that takes 14 days, because the method incubates for 14 days. Every number comes with its raw data, so a reader can apply Finnrick’s thresholds, or their own, to the measured values. The report is a signed record you can check at hkanalytical.com/verify. See the example report.