The 2005 oxytocin trust study did not test love, relationship repair, sexual desire, or mental health care. Michael Kosfeld and colleagues in Nature published the experiment in June 2005. It tested a brief money game in healthy male college students.
Men given intranasal oxytocin sent more experimental cash to a stranger than men given a placebo. That single paper helped popularize the "trust hormone" and "liquid trust" slogans.
Yet the trial measured only four anonymous money choices in a Zurich lab. Over the next twenty years, larger follow-up studies struggled to repeat that simple result.
What the 2005 oxytocin trust study actually tested
Kosfeld's team asked whether intranasal oxytocin changes how much money a person entrusts to an anonymous stranger. They also checked whether that change required a human partner rather than a willingness to take financial risk.
The study enrolled 194 healthy male students in Zurich. Their average age was 22.0 years (standard deviation, or typical spread, 3.4 years). Initially, 128 men entered the trust experiment and 66 entered a separate risk-control experiment.
Researchers excluded men with significant medical or psychiatric illness, medication use, substance abuse, or smoking over 15 cigarettes daily. Sixteen men were excluded before analysis. Twelve had incorrect dose administration (7 trust, 5 risk, including one failed task check), and 4 doubted their partner was human.
In this randomized, double-blind trial, neither students nor staff knew who received oxytocin or placebo. Men received a single intranasal dose of 24 international units (IU) of oxytocin or matched placebo 50 minutes before the task. A five-minute waiting period permitted minimal social contact. These protocol details are study facts, not a dosing instruction or a description of any Promise product.
Anonymous pairs started each round with 12 monetary units (MU). The investor chose to send 0, 4, 8, or 12 MU to the trustee. Any sent money was tripled, and the trustee could return some or none of it.
Each investor played four rounds with different random partners and received no feedback between rounds. The investor's four-round average transfer served as the behavioral measure of trust. The paper reported no trial registry identifier or prospectively registered primary endpoint.
For broader hormone background and receptor biology, see our guide to the oxytocin peptide. Oxytocin is coming soon to Promise as a compounded medication, which is different from an FDA-approved product: the formulation Promise will offer is not FDA-approved.
The headline numbers, risk control, and side effects
Among 58 analyzed investors, 13 of 29 on oxytocin (45%, or nearly half) chose the maximal average transfer. Only 6 of 29 on placebo (21%, or about 1 in 5) did the same. The paper states that "[t]he investors' average transfer is 17% higher in the oxytocin group" (9.6 versus 8.1 MU).
That comparison used a one-sided Mann–Whitney test (P = 0.029), which checks for a shift in one expected direction only. An appended printing correction (corrigendum) notes that Figure 2a omitted three placebo observations at 10 MU. The notice states that "[a]ll statistical tests in the paper were done including these observations" already.
In the separate risk-control game, a random mechanism replaced the human trustee with identical financial odds. Both groups averaged 7.5 MU (two-sided P = 0.983), which the authors read as evidence against a general rise in gambling.
Trusting a stranger and being trustworthy were separate outcomes. Oxytocin did not significantly raise trustee back-transfers (P > 0.243 at each positive transfer level) or alter return expectations. The authors' betrayal-aversion explanation remained an unmeasured behavioral interpretation.
| Measure | Oxytocin | Placebo |
|---|---|---|
| Trust investors analyzed | 29 | 29 |
| Mean average transfer, MU | 9.6 | 8.1 |
| Median average transfer, MU | 10 | 8 |
| Standard deviation, MU | 2.8 | 3.1 |
| Investors with maximal average transfer | 13 (45%) | 6 (21%) |
| Risk-control participants analyzed | 31 | 30 |
| Risk-control mean average transfer, MU | 7.5 | 7.5 |
| Risk-control median average transfer, MU | 8 | 8 |
| Risk-control standard deviation, MU | 3.3 | 3.4 |
The 2005 paper did not report adverse-event counts by arm. Its mood and calmness questionnaires do not establish clinical safety.
| Safety reporting available for this trial | What can be stated |
|---|---|
| Adverse-event incidence by arm | Not reported in the original paper |
| Serious adverse-event incidence by arm | Not reported in the original paper |
| Long-term/repeated-use safety | Not measured by this brief behavioral experiment |
Does oxytocin increase trust in replication studies?
In a 2015 Perspectives on Psychological Science review, Gideon Nave, Colin Camerer, and Michael McCullough pooled seven intranasal trust studies (481 participants). The combined effect was not reliably different from zero (P = 0.45). The standardized effect size was tiny (Cohen's d = 0.077, 95% CI −0.124 to 0.278).
Their review also flagged flaws in blood-hormone and genetic claims. Separately, a December 2005 Hormones and Behavior study by Paul Zak's team measured blood oxytocin around trustworthy behavior. That blood study did not test intranasal oxytocin administration.
In a 2016 Biological Psychiatry paper, Hasse Walum, Irwin Waldman, and Larry Young analyzed historical sample sizes across the field. They estimated 16% statistical power (the chance of detecting a true effect) in typical healthy-volunteer studies and 12% in clinical trials. Low power can miss real effects and exaggerate small positive findings.
Carolyn Declerck, Ernst Fehr, and colleagues then ran a 2020 registered replication in Nature Human Behaviour with over 95% stated power. Oxytocin showed no effect in the main minimal-social-contact condition, while an exploratory no-contact trend in low-trusting men required confirmation. A failed replication weakens confidence in that tested effect without proving oxytocin has no biological role.
A May 2026 Cortex registered report by Charlotte Kroll's team (online March 11, 2026) found no trust increase in 211 participants. Pooling those data with 321 Declerck observations (n = 532) also showed no increase. The pooled test supported equivalence within a minimal effect range, though the new sample alone was inconclusive.
As of October 17, 2026, the day this article was written, the newest trial is an August 6, 2026 PNAS study. Bodo Vogt, Paul Bengart, Carolyn Declerck, and Ernst Fehr studied 359 men selected for low dispositional trust. Oxytocin increased trusting transfers by roughly 15%, and by 16.9% when pooled with an earlier sample (n = 219).
An August 12, 2026 University of Zurich release noted that biological mechanisms were not directly tested and applicability to women needs study.
How a lab game became 'love hormone' marketing
Animal attachment and reproductive research also predated the 2005 trust study. Our 1984 Nobel Prize peptide synthesis article mentions Vincent du Vigneaud's earlier oxytocin synthesis.
After 2005, online pitches stretched nicknames like "love hormone," "liquid trust," and "moral molecule" into broad social claims. Yet four anonymous money transfers do not show that a dose automatically creates affection.
A brief student money game also does not prove relationship repair or sexual benefit. Likewise, hospital childbirth indications do not validate social-use products.
PT-141, or bremelanotide, belongs to a separate sexual-health category rather than the trust research discussed here. Brand-name Vyleesi (revised 3/2024) is indicated only for premenopausal women with acquired, generalized hypoactive sexual desire disorder (HSDD, or distressing low desire). Its label states that its mechanism in HSDD is unknown and it is not indicated to enhance sexual performance.
PT-141 is dispensed by a licensed U.S. compounding pharmacy as a compounded medication, which is different from an FDA-approved product: the formulation offered here is not FDA-approved.
Pitocin's obstetric label and how compounded blends differ
The Pitocin label (code R03/2026, published May 21, 2026) covers an obstetric injection. It is indicated "for the initiation or improvement of uterine contractions" when suitable to achieve vaginal delivery. It "is not indicated for elective induction of labor," while postpartum use targets third-stage contractions and bleeding.
For labor induction or augmentation, Warnings require intravenous administration with adequate medical supervision in a hospital. Contraindications list specific antepartum exclusions, and Adverse Reactions note uterine spasm or rupture risks. Precautions also describe an intrinsic antidiuretic (water-retaining) effect.
The label reports severe water intoxication with convulsions and coma associated with a slow 24-hour infusion. It separately reports maternal death due to oxytocin-induced water intoxication. Those hospital warnings do not give side-effect rates for intranasal experiments, and trust studies do not support pregnancy use.
As the FDA notes on compounded drugs, compounded medications are not FDA-approved. The agency does not review their safety, effectiveness, or quality before marketing, though compounded drugs can serve an important medical need.
Separately, the PT-141 / Oxytocin blend combines both peptides in one listing. It is coming soon to Promise as a compounded medication, which is different from an FDA-approved product: the formulation Promise will offer is not FDA-approved.
Single-ingredient trials do not prove blend benefits. Vyleesi is contraindicated in uncontrolled hypertension or known cardiovascular disease. Its label tells patients to contact their healthcare provider for persistent or severe nausea.
What the study means for a telehealth visit
None of the trust-game trials tested a Promise product or blend. A one-day student money game cannot stand in for a different product, route, device, population, or long-term goal.
A licensed provider may still prescribe — that decision is between you and your doctor. At Promise, a licensed provider reviews every request and not everyone qualifies. Any medication prescribed through Promise is dispensed by a licensed U.S. compounding pharmacy.
