General information about published U.S. regulations and peer-reviewed literature — not legal, medical, or manufacturing advice.

A peptide active ingredient has no endotoxin limit of its own. The number on a certificate of analysis means something only once you know the formula it came from, whether the method was shown to work in that peptide's own matrix, and whether the run's control recovered. As of March 2026, FDA also no longer writes the test as horseshoe-crab lysate by default.

What is the endotoxin limit for a peptide API?

There is no fixed figure for an active ingredient in isolation — the limit is calculated. FDA states the endotoxin limit is K/M, where "K is 5.0 EU/kilogram (kg.), which represents the approximate threshold pyrogen dose for humans and rabbits," and "If a product is labeled for intrathecal injection, then K is 0.2 EU/kg," per FDA Inspection Technical Guide No. 40. The peer-reviewed literature states the same maxima — "5 EU/kg/h for most drug products and 0.2 EU/kg/h for intrathecally administered drugs" — in Dobrovolskaia et al. (PMC2900157).

M is "the rabbit pyrogen test dose or the maximum human dose per kilogram that would be administered in a single one hour period, whichever is larger," in the same guide — so the finished product sets the number the same powder has to meet. Water carries its own fixed limit: Water for Injection, Sterile Water for Injection and Sterile Water for Irrigation have "an allowable endotoxin limit of 0.25 Endotoxin Units (EU)/ml."

One sourcing caveat: ITG No. 40 is dated 3/20/85, so while its arithmetic is corroborated above, the document of record for current testing recommendations is FDA's Pyrogen and Endotoxins Testing: Questions and Answers. A CoA line reading only "endotoxin: complies" is unverifiable unless the limit and its basis travel with it, so ask for the method beside the result. Our quality documentation covers what a Cornerstone Peptide CoA reports.

Which methods satisfy USP <85>?

Three technique families are in use — gel-clot, turbidimetric, and chromogenic — per Dobrovolskaia et al. (PMC2900157). FDA describes the documents its guidance addresses, including USP Chapter <85>, "Bacterial Endotoxins Test," as covering "gel clot, photometric, and kinetic test methods" (guidance page).

The assay's floor is a reagent property, not a choice. Lambda (λ) is "the sensitivity of the lysate provided for each lot of the lysate by the manufacturer" — PMC2900157 — and it fixes the lowest concentration a gel-clot test can detect. One published gel-clot method used a reagent whose "sensitivity … contained in each vial was 0.125 EU/ml," scoring a tube positive only if the gel "remained intact in the bottom of the reaction tube after an inversion of 180°" — Sharma et al. (PMC3543581). Because λ sets that floor, a sample may be diluted to beat interference only as far as the maximum valid dilution, the ceiling set out in FDA Inspection Technical Guide No. 40; past that the test can no longer see the limit it enforces, so the dilution belongs on the record with the result. For Cornerstone Peptide catalog items, LAL endotoxin testing is available on request; see capabilities for how production is run.

Do you still have to use horseshoe-crab LAL in 2026?

No — that changed this year. FDA's Pyrogen and Endotoxins Testing Q&A is now a Final Level 2 Revised Guidance dated March 2026 (docket FDA-2013-S-0610), and in it FDA "removed certain references to Limulus Amoebocyte Lysate (LAL) testing to accommodate a broader scope of recombinant reagents," with users of those reagents expected to "verify that the assay method is suitable for its intended purpose" — FDA, March 18, 2026. That same announcement ties the change to USP Chapter <86>, "Bacterial Endotoxins Test Using Recombinant Reagents," which presents non-animal-derived alternatives; what is cited here is FDA's characterization of that chapter, not the compendial text itself. Read the change precisely: flexibility plus a suitability obligation, not a mandate to switch and not one chapter displacing the other.

Supply is why this matters commercially. "Procuring blood for LAL testing involves capturing and bleeding over 500,000 crabs from wild marine populations each year," with biomedical use rising "from 335,501 crabs in 2004 to 575,760 crabs in 2017," per Gorman (PMC7612741). That is the pressure behind the reagent question, and it is why a buyer may see either reagent class named on a peptide CoA — which makes the suitability verification above, not the reagent's origin, the line that decides whether the number holds.

Why does a clean peptide fail — or falsely pass — the test?

Because the peptide is part of the assay, not just its subject. FDA's inspection guidance states that "The necessity to validate the reliability and accuracy of the LAL method for each product tested cannot be over-emphasized," and names the mechanisms: "Chemical inhibitors cause chelation of divalent cations necessary for the LAL reaction," while "Physical inhibitors include adsorption of endotoxin, or product viscosity" — ITG No. 40. In FDA's inspectional experience with finished products, where a pyrogen problem traced back to a raw material, "it was the active drug substance" — the API, not the water or the containers.

That is what the run's control is for: a result is valid only if the spiked positive product control recovers 50% to 200% of the added endotoxin, and outside that range the number cannot be reported (PMC2900157; Reich et al. (PMC6412962)). The harder failure mode is time-dependent. Reich et al. describe Low Endotoxin Recovery as "the inability to recover more than 50% activity over time when endotoxin is added to an undiluted product," and report from prior work that the masking is "driven by" "a simultaneous presence of a surfactant (e.g., Polysorbate 20 or Polysorbate 80) and a chelator (e.g., sodium citrate)." Because the definition turns on activity lost *over time*, the evidence that speaks to it is a hold-time study rather than a time-zero result. Separately, the same authors state that "regulatory authorities request hold-time studies of endotoxin in addition to pharmacopoeial requirements"; their own analysis covered hold times up to 7 days.

Can you rely on a supplier's certificate of analysis?

Only in part, and the regulation says which part. Under 21 CFR § 211.84(d)(2), a drug-product manufacturer may accept a supplier's report of analysis for a component only if it also performs at least one specific identity test itself and "establishes the reliability of the supplier's analyses through appropriate validation of the supplier's test results at appropriate intervals." Section 211.84(d)(6) adds that "Each lot of a component … with potential for microbiological contamination that is objectionable in view of its intended use shall be subjected to microbiological tests before use."

Two further provisions decide whether an endotoxin number is defensible at all. 21 CFR § 211.165(e) requires that "The accuracy, sensitivity, specificity, and reproducibility of test methods employed by the firm shall be established and documented," and 21 CFR § 211.194(a)(2) requires that "The suitability of all testing methods used shall be verified under actual conditions of use" — in the peptide matrix, not in water. The batch obligation sits with the drug product: 21 CFR § 211.167(a) requires "appropriate laboratory testing" for each batch "purporting to be sterile and/or pyrogen-free." Part 211 is drug-product CGMP, so these are the buyer's obligations a supplier's paperwork has to support — a boundary covered further in how U.S. law classifies and regulates peptides and what the regulations require of a peptide synthesis company. The animal test survives elsewhere: it remains codified for licensed biologics at 21 CFR § 610.13(b) — part 610, biologics, not peptide APIs.

In practice that is a short list of what should arrive with a lot: the technique used, λ or the equivalent sensitivity claim, the dilution run, the positive-product-control recovery, and hold-time data supporting the method in that matrix. Endotoxin testing on the Cornerstone Peptide catalog is by request, so ask for that list when you request it — and tell us the finished-product context so the limit can be set against your product.

Frequently asked questions

What is the endotoxin limit for a peptide API? There is no fixed number for the ingredient alone — it is K/M, with K at 5.0 EU/kg (0.2 EU/kg for intrathecal labeling) and M derived from the finished product, per FDA ITG No. 40.

Which test techniques satisfy USP <85>? Gel-clot, turbidimetric, and chromogenic (PMC2900157) — with FDA describing the standards it addresses as covering "gel clot, photometric, and kinetic test methods" (guidance page).

Is horseshoe-crab LAL still required in 2026? No — FDA's March 2026 revision "removed certain references to Limulus Amoebocyte Lysate (LAL) testing to accommodate a broader scope of recombinant reagents," with users of those reagents expected to "verify that the assay method is suitable for its intended purpose" (FDA, March 18, 2026).

Can a buyer rely on a supplier's endotoxin result? Only in part: 21 CFR § 211.84(d)(2) still requires one specific identity test plus validation of the supplier's results at appropriate intervals, and § 211.194(a)(2) requires suitability verified under actual conditions of use.

Why would a clean peptide fail or falsely pass? Interference and masking — chelation, adsorption, and viscosity are named as inhibitors in ITG No. 40, and Low Endotoxin Recovery (a surfactant plus a chelator masking added endotoxin below 50% recovery over time) in Reich et al. (PMC6412962).

When is an endotoxin result invalid? When the spiked positive product control recovers below 50% or above 200% of the added endotoxin — the run is invalid and the number cannot be reported (PMC6412962).