I went looking for the paper trail behind the LL-37 dosing charts. There isn’t one. Numbers tagged [P#] below trace to real papers, listed at the bottom. Check my work.
Here’s the setup. Peptide forums and vendor pages hand out LL-37 doses to the microgram. Cycle lengths. Frequencies. Stated like court testimony. I wanted to see the source document behind those numbers. I couldn’t find one, because it doesn’t exist. There is no dose-finding trial in humans for injecting LL-37 to treat what it’s marketed for. Every confident protocol you’ve seen is a guess dressed up in a lab coat.
I’m not handing you a number either. What I can hand you is the trail: where those numbers actually came from, why the margin for error is thinner than anyone selling you a vial wants to admit, and what to do if you’re going ahead regardless. Call it harm reduction. Don’t call it a recipe.
The paper trail runs cold
A real dosing source points at a study and says: this much, this often, tested and confirmed. For injected LL-37, that document isn’t in the file.
What is in the file: a 2014 randomized, placebo-controlled trial in Wound Repair and Regeneration, 34 patients, venous leg ulcers, lower doses applied topically speeding healing without trouble [P4]. Solid work. Wrong case entirely. That was a gel on an open wound, dosed by surface concentration. Not a needle under skin, not a microgram figure. The other lead is an early-phase cancer study, NCT02225366, injecting LL-37 straight into melanoma tumors under medical supervision [P8]. Different patient, different intent, different math.
So when I tell you I won’t give you a number, that’s not me stonewalling. Giving you a fake-precise figure would be the lie. Anyone handing you one is either guessing or copying somebody else’s guess. You should know that before a needle enters the picture.
The switch nobody flags
Here’s where the file gets slippery, and where the forum charts quietly change the subject on you. The real human evidence talks in concentrations, the level measured in a dish or laid across a wound surface. What you get told to inject is a different animal: an amount pushed under your skin that then spreads, gets chewed up by your own enzymes, and lands at some unknown concentration near whatever cells it happens to reach.
Those aren’t the same unit. You can’t convert one to the other on a bar napkin.
Why it matters: the toxicity data I’m about to show you is stated in concentration terms, the point where LL-37 starts damaging your own cells. A microgram number injected under skin gives you no way to check whether you’re staying under that line, because the molecule is unstable and scatters unpredictably [P3]. Even the careful ones, the “start low” crowd, are dosing blind. Their number was never tied to what their tissue will actually see.
Following the number back to its source
I traced the circulation pattern, and it’s the oldest trick in the book. Someone posts a protocol. A vendor lifts it onto a product page for the appearance of authority. A forum quotes the vendor. Round and round until repetition has done the work evidence should have done.
Try this yourself. Take any LL-37 protocol you’ve run across and ask for the human study behind the number. For these uses, there isn’t one. What you’ll get instead is “this is what people use.” That’s the circular sourcing, confirmed. A dose that only traces back to other people using it isn’t evidence-backed. It’s a habit with decent marketing behind it. Knowing that is the first thing that’ll protect you.
The one thing that holds: the margin is thin
Here’s the fact that survives every round of scrutiny, the one piece of this file I trust completely. LL-37 doesn’t have a wide gap between “working” and “hurting you.” The mechanism that lets it tear open bacterial membranes doesn’t stop politely at the bacteria.
A 2013 review in Frontiers in Immunology says it plainly: helpful at low concentrations, toxic to human cells above roughly 10 micromolar. Narrow window, their words [P3]. A 2025 review in the International Journal of Molecular Sciences backs it from another angle, calling native LL-37 cytotoxic and unstable, which is exactly why researchers are trying to redesign it instead of using it straight [P7].
Sit with that. Most compounds, too much is wasted, not dangerous. LL-37 doesn’t work that way. Too much moves you out of helpful and into the zone where the peptide turns on your own cells. A forum chart offers zero protection against that, because whoever wrote it wasn’t measuring your cells, they were measuring nothing at all.
And it’s unstable on top of that. Your own enzymes and bacterial enzymes break it down fast [P3]. So even the link between what you inject and what actually does anything is unreliable. You’re not dosing a well-behaved drug. You’re dosing a fragile peptide with a thin safety margin, off numbers nobody ever checked. Three strikes.
Exhibit A: what “too much” looked like in a real patient
I don’t deal in hypotheticals when there’s an actual case on file. A 2018 report in the Journal of Cutaneous Pathology documents a melanoma patient who developed new skin lesions, some reading as skin cancer under the microscope, after weeks of LL-37 injected into tumors. The lesions cleared within roughly two months of stopping [P5].
Read it straight: one case, cancer-treatment context, intratumoral dosing. Not proof this happens to everyone injecting LL-37 under skin. But it is a documented, reversible adverse reaction in a real human, and it’s exactly the kind of red flag most dosing pages leave out of the file entirely.
There’s a wider concern too. The 2025 review notes LL-37 can act as an autoantigen, something that can provoke the immune system into attacking the body, and it’s been tied to autoimmune and inflammatory conditions [P7]. Deliberately pushing your LL-37 levels up with an unvalidated injected dose isn’t a free move against that backdrop. No certainty of disaster here. Just real downside against unproven upside, which is a bad trade for a decision you’re making alone at your kitchen table.
If you’re going ahead anyway, here’s the harm-reduction version
Read this even if you’ve made up your mind. If you’re using LL-37 regardless of everything above, don’t freelance the number by yourself. The single move that cuts your risk the most: take the dosing decision off the forum and put it on someone licensed who has to answer for it. That’s also the only lawful route.
That’s the model FormBlends runs on, and it’s the reason I rank it first. A clinician reviews your history, your other medications, your actual conditions. Decides whether LL-37 belongs anywhere near your situation. Owns the dose, and any change to it later, instead of leaving that call to you. Prescribes only when there’s a reason to, then a licensed pharmacy compounds it. A named person is on the hook with you. That’s the opposite of guessing alone.
If you’re comparing options, HealthRX runs the same accountable-clinician model and earns the second spot for the same reason.
A few more things I’d put in writing before you touch a needle:
- Don’t invent a number off a product page. The microgram figures in circulation are guesses, not trial-derived doses, and the safe window is narrow [P3].
- Assume less beats more, because the gap between helpful and cell-damaging is small [P3][P7]. That’s an argument for a clinician setting the dose, not for picking a low number yourself and crossing your fingers.
- Keep a log. Every dose, every reaction, dated. A fuzzy memory is worthless at a follow-up appointment. A tool like the FormBlends tracker app exists to record exactly this, dose by dose, side effect by side effect, building a real timeline a clinician can actually use. It’s a logging tool. Nothing more, nothing to check out.
- Watch your skin specifically. The one documented human adverse reaction to injected LL-37 was dermatologic, and it only reversed after stopping [P5]. New or worsening lesions mean stop and get seen, not push through.
- “Natural” doesn’t mean safe to escalate. Your body releases LL-37 in tiny, controlled amounts at specific sites. Injecting a dose is a different event altogether. The antimicrobial and biofilm effects people cite were measured in lab dishes at controlled concentrations [P2], not by someone eyeballing a syringe at home.
The call
There’s no research-backed dose for injecting LL-37 for the things it’s sold to fix. The numbers making the rounds are repeated guesses. The safe window is genuinely thin. The molecule is unstable. And there’s a documented case of it causing harm in a real person.
If there’s one line to carry out of this file, it’s this: the dose was never the thing protecting you. Whether a clinician is in the room is. Buy the supervision. Log everything. Treat every confident microgram figure like a claim that hasn’t been verified yet, because it hasn’t.
Questions I keep getting asked
What’s the correct dose of LL-37 to inject?
There isn’t one. None has been established for injecting LL-37 for the conditions it’s marketed against. No human dose-finding trial exists for subcutaneous LL-37 in chronic infection, gut health, or immune support [P4][P8]. Any microgram figure you’re handed is a circulated convention, not a study result.
Why do vendors and forums post such precise doses if no trial backs them?
Precision is a performance, not evidence. A protocol gets posted, a vendor copies it to look credible, another forum quotes the vendor, and the guess hardens into a “standard” purely through repetition. Ask anyone for the human study behind their number. For these uses, there isn’t one to hand you.
Does a higher dose work better, or does it just mean more risk?
More risk, not more benefit. LL-37 has a thin margin between its active range and the level where it damages human cells, roughly above 10 micromolar in lab work [P3][P7]. The same membrane-tearing action that kills microbes doesn’t spare your own cells once concentration climbs. A little too much can cross that line.
Has injected LL-37 actually hurt someone?
Yes, on record. A 2018 case report documented a melanoma patient who developed multiple new skin lesions, some resembling skin cancer under the microscope, after weeks of LL-37 injected into tumors. The lesions cleared within about two months of stopping [P5]. One case, cancer-treatment setting, so it’s not a universal prediction. But it’s a real, reversible adverse reaction in a human being.
If I’m using it regardless, what actually lowers my risk?
Getting a licensed clinician to review your history and medications, decide if LL-37 has any place in your case, and own the dose instead of leaving you with a forum chart. That’s also the only lawful path. Keep a written log of every dose and reaction. Watch specifically for new or worsening skin changes, the one documented human adverse signal [P5]. And don’t let the word “natural” fool you once you’re injecting amounts your body never produces on its own.
What is LL-37, and where does it come from?
Thirty-seven amino acids your own body already makes, mostly in skin cells, neutrophils, and airway linings. Part of a family called cathelicidins, sitting at the front line of innate immunity. It multitasks: rips open bacterial membranes, signals immune cells, may help tissue repair. The synthetic version tries to copy that same sequence.
Is buying and using LL-37 legal?
In the U.S., LL-37 has no FDA approval as a drug, so it can’t legally be sold as a finished pharmaceutical for human use. Research-chemical outfits sell it “for lab use only,” which is a gray zone by design. The cleaner route, if a clinician actually thinks it fits your care, runs through a licensed compounding pharmacy under physician oversight, like FormBlends, where the sourcing and dosing get documented and somebody’s name is on it.
What side effects are actually documented?
Thin data. Mostly in-vitro work and a handful of early-phase trials, largely topical wound studies. Reported issues include local irritation at injection sites, and at higher lab concentrations the peptide turns cytotoxic to host cells, not just bacteria. Systemic effects in humans aren’t well mapped. Anyone claiming a clean safety profile at therapeutic doses is ahead of what the evidence actually says.
Does LL-37 actually deliver on what people claim?
Promising, not settled. Cell and animal studies show genuine antimicrobial and immune-modulating activity. Human trial data is thin, mostly small wound-healing studies plus a few respiratory looks, results mixed. No large randomized trial has nailed down a clear clinical benefit for any use yet. The gap between petri dish and person is still wide.
References
[P2] Overhage J, Campisano A, Bains M, et al. Human host defense peptide LL-37 prevents bacterial biofilm formation. Infection and Immunity. 2008;76(9):4176-4182. https://pubmed.ncbi.nlm.nih.gov/18591225/
[P3] Ramos R, Silva JP, Rodrigues AC, et al. Wound healing activity of the human antimicrobial peptide LL37. Peptides. 2011;32(7):1469-1476. https://pubmed.ncbi.nlm.nih.gov/21672586/
[P4] Grönberg A, Mahlapuu M, Ståhle M, Whately-Smith C, Rollman O. Treatment with LL-37 is safe and effective in enhancing healing of hard-to-heal venous leg ulcers: a randomized, placebo-controlled clinical trial. Wound Repair and Regeneration. 2014;22(5):613-621.
[P5] Dermatologic toxicity from novel therapy using antimicrobial peptide LL-37 in melanoma: a detailed examination of the clinicopathologic features. Journal of Cutaneous Pathology. 2018;45(7):524-530.
[P7] Antimicrobial peptides of the cathelicidin family: focus on LL-37 and its modifications. International Journal of Molecular Sciences. 2025.
