DSIP is sold as a sleep peptide, and the dose figure attached to it is unusually traceable: it comes from a small set of human studies published between 1984 and 1993, most of them by one researcher. This page opens those studies, reports what they administered, and sets out three things about the figure that the secondary sources drop.
This is a report of what published sources state. It is not a protocol, and nothing here is a recommendation.
What the molecule is
DSIP is a nonapeptide — nine amino acids, Trp-Ala-Gly-Gly-Asp-Ala-Ser-Gly-Glu. PubChem lists it as CID 68816, formula C35H48N10O15, molecular weight 848.8 g/mol. It was isolated in the 1970s from the cerebral venous blood of rabbits in induced slow-wave sleep, which is where the name comes from.
FDA's compounding records use a different name for it: emideltide. Anyone searching regulatory databases for "DSIP" and finding nothing is searching the wrong string, which is a small point that accounts for a good deal of confusion about its status.
The literature, counted
A PubMed search for the exact phrase in titles and abstracts returned 407 records on 13 September 2026, of which 97 are human-tagged. By the standards of this catalogue that is a deep file — most research peptides return single-digit or low-double-digit human counts.
The distribution is what makes it unusual:
| Period | Records | Share |
|---|---|---|
| 1977–1990 | 183 | 45% |
| 1991–2009 | 194 | 48% |
| 2010–2026 | 30 | 7% |
ClinicalTrials.gov returned zero registrations for either name. DailyMed returned zero records. So the position is: a compound with a substantial historical literature, no registered trial anywhere, and no approved product.
That shape is worth naming plainly, because it is the opposite of how a research peptide's evidence base is usually described. This is not an emerging compound whose trials are pending. It is a compound that was worked on hard for about fifteen years, produced equivocal results, and was set down. The research market picked it up afterwards.
The dose figures, and where each one comes from
Four human studies were opened for this page. Every one used the intravenous route.
| Study | Population | Dose and route | Result |
|---|---|---|---|
| Schneider-Helmert, Eur Neurol 1984 | Insomnia patients, two studies summarised | 25 nmol/kg body weight, injection before sleep | Improved sleep after single injections; build-up with repeated administration, sleep structure normalising after about four |
| Schneider-Helmert, Eur Neurol 1987 | 14 middle-aged chronic insomniacs | Intravenous, 7 successive nights, double-blind placebo-controlled | Substantially improved night sleep; daytime alertness and performance rose significantly; effects held into the first placebo post-treatment night |
| Bes et al., Neuropsychobiology 1992 | 16 chronic insomniacs, matched-pairs parallel groups | 25 nmol/kg intravenously, afternoons before nights 3, 4 and 5 | Higher sleep efficiency and shorter latency than placebo, but the authors judged the effects weak and partly an artefact of the placebo group; no change in subjective sleep quality |
| Giusti et al., Psychoneuroendocrinology 1993 | 8 healthy women aged 17–36 | 25 µg/kg intravenously over 30 minutes; and 25 µg/kg infused 21:30–22:30 | No effect on growth hormone or prolactin, basal, arginine-stimulated or circadian |
The unit problem
Two of these studies report "25" and they do not mean the same quantity.
- 25 nmol/kg × 848.8 g/mol = 21.2 µg/kg — about 1.48 mg for a 70 kg adult.
- 25 µg/kg = 25 µg/kg — about 1.75 mg for a 70 kg adult.
The two differ by roughly 18 percent. That is close enough that a figure with its unit dropped still looks reasonable, and far enough apart to matter. A secondary source quoting "25" for DSIP without the unit is not a usable source, and there are a great many of them.
The route problem
This is the larger one. Every human administration above went into a vein — bolus or timed infusion, in a clinical setting, generally with polysomnography recording the result.
The research market sells DSIP as a lyophilised powder in multi-milligram vials, reconstituted and injected subcutaneously. No published human study opened for this page did that. A milligram figure derived from intravenous work does not transfer to a subcutaneous depot: the peak concentration, the time to it and the shape of the exposure curve are all different, and for a peptide whose reported sleep effect was tied to timing before sleep, that is not a detail.
The general problem of reading a figure across formulations and routes is covered in our guide to how to read a dosing claim.
The result the marketing does not carry
Schneider-Helmert's 1987 study is the one every retail page cites, usually as "clinically proven". The 1992 study by Bes and colleagues used the same compound, the same 25 nmol/kg, the same intravenous route and a double-blind design in a comparable population, and reached the opposite conclusion — that the effects were weak, that one measure of subjective tiredness moved and subjective sleep quality did not, and that short-term treatment was "not likely to be of major therapeutic benefit."
Two double-blind studies, same dose, opposite readings, and no registered trial in the thirty-three years since to break the tie. That is the accurate summary of the human sleep evidence for DSIP, and it is not what the compound is sold on.
The one substance the committee declined
On 23-24 July 2026 FDA's Pharmacy Compounding Advisory Committee reviewed seven nominated peptides for the 503A bulks list — the list that would permit a pharmacy to compound from them. It recommended six. Emideltide was the one it voted against, failing 6 to 7 with one abstention, while semax passed 8-5, epitalon 7-5 and TB-500 8-6.
The vote is advisory and FDA has not acted on it, so it changes nothing about the compound's formal status. It is worth recording because of what the committee was looking at: a compound whose human file is larger than most of the others in the room, and whose two double-blind sleep studies point in opposite directions. A deeper literature that fails to converge is not a stronger case than a thin one, and on this occasion it was treated as a weaker one. The vote and what it did and did not do is set out on our page on cognitive peptides.
One safety line worth reporting
The 1993 endocrine paper contains an aside that is easy to miss: the authors describe the doses they studied as ones "which are known to modify ECG patterns." They cite it as context for their own null result rather than reporting a cardiac finding of their own, and this page can add nothing to it beyond noting that it is there, in a peer-reviewed paper, attached to the dose range under discussion.
What can be said about the adverse-event record is that there is essentially none to read. Our count of FDA's adverse event database sets out why it holds almost nothing for compounds in this catalogue, and why that absence is a fact about reporting channels rather than about safety.
What is not established
- No approved indication, anywhere in the US label database. No labelled dose, no reviewed route, no approved formulation.
- No registered trial. Zero on ClinicalTrials.gov under either name, so nothing is pending that would resolve the 1987-versus-1992 disagreement.
- No subcutaneous human data in the studies opened here, which is the route the sold product is intended for.
- No published human pharmacokinetic profile that this page was able to open, which is why no half-life figure appears on it.
- No long-term safety data. The longest human exposure in these studies is seven consecutive nights.
What a reader can check in five minutes
Search PubMed for "delta sleep-inducing peptide"[tiab] and read the record count, then add AND 1977:1990[dp] and compare. The 45 percent figure above falls straight out. Search ClinicalTrials.gov for the same phrase and for "emideltide" and confirm both return nothing. Then open PMID 3622582 and PMID 1299794 side by side — the 1987 and 1992 studies — and read the two conclusions in sequence.
Related on this site
- How to read a dosing claim — units, routes, formulations and the errors that survive transcription
- FDA's category 2 list, indexed — where emideltide sits in FDA's compounding records
- What the adverse-event record contains — and why near-zero counts prove nothing
- How to reconstitute peptides — the mechanics of getting from a lyophilised vial to a concentration
Sources and dates
- PubChem CID 68816, delta sleep-inducing peptide — formula and molecular weight. Read 13 September 2026.
- PubMed record counts via NCBI E-utilities, exact-phrase title/abstract searches. Run 13 September 2026.
- ClinicalTrials.gov API v2, queries for "delta sleep-inducing peptide" and "emideltide". Run 13 September 2026, both returning zero.
- DailyMed SPL database, queries for delta sleep-inducing peptide, DSIP and emideltide. Run 13 September 2026, all returning zero.
- Schneider-Helmert D. DSIP in insomnia. Eur Neurol. 1984;23(5):358–63. PMID 6391925.
- Schneider-Helmert D. Effects of delta-sleep-inducing peptide on 24-hour sleep-wake behaviour in severe chronic insomnia. Eur Neurol. 1987;27(2):120–9. PMID 3622582.
- Bes F, Hofman W, Schuur J, Van Boxtel C. Effects of delta sleep-inducing peptide on sleep of chronic insomniac patients. A double-blind study. Neuropsychobiology. 1992;26(4):193–7. PMID 1299794.
- Giusti M, et al. Delta sleep-inducing peptide administration does not influence growth hormone and prolactin secretion in normal women. Psychoneuroendocrinology. 1993;18(1):79–84. PMID 8475226.
- FDA, Certain Bulk Drug Substances for Use in Compounding That May Present Significant Safety Risks, nominated-but-withdrawn section. Content current as of 22 April 2026; read 13 September 2026.