Intranasal oxytocin: anatomy of a replication crisis

How it started

Oxytocin is a peptide hormone made in the hypothalamus and released from the posterior pituitary; its classic actions are contracting the uterus during labor and ejecting milk during breastfeeding¹³. In the decade before 2016 numerous studies associated nasal oxytocin with various aspects of human social behavior⁴. The peptide entered popular culture as the "love hormone" and the clinic as a possible autism treatment; it came to be given in clinical practice to many children with autism¹. One of the most cited papers, published in Nature in 2005, reported that intranasal oxytocin substantially increased trust in an economic game⁵.

What happened with the large trials

• Childhood autism (NEJM 2021): 290 children and adolescents, 24 weeks. Difference on the social-withdrawal subscale (ABC-mSW) vs placebo of −0.2 (95% CI −1.5 to 1.0), P = 0.61¹.
• Adult autism (2020): 106 adults aged 18 to 48, 6 weeks; no difference vs placebo on the primary outcome (effect size −0.08; P = 0.69). Among secondary outcomes, repetitive behavior was reduced².
• Obesity (NEJM Evidence 2024): 61 adults, 8 weeks. Weight change of 0.20 vs 0.26 kg, P = 0.934; oxytocin was associated with lower caloric intake at a test meal³.
• Autism in young children (2023): 87 children aged 3 to 12, 12 weeks; no overall effect on the SRS-2 social responsiveness scale (p = 0.686), with a possible signal in the 3-to-5-year subgroup that the authors say needs confirmation in new trials¹¹.
• Trust (registered replication, 2020): a double-blind study with more than 95% statistical power found no effect of oxytocin on trust in the condition that reproduced the 2005 experiment⁶.
• Trust (Cortex 2026): a new replication with 211 participants, analyzed together with the 2020 data (532 in total), found no evidence that intranasal oxytocin increases trust⁷.

Why the small studies failed

In 2016 a statistical analysis of the literature concluded there was a "high probability that most of the published intranasal OT findings do not represent true effects"⁴. The authors note that these studies are generally underpowered, and also weigh low pre-study odds and bias⁴. This is not unique to oxytocin: a 2013 review showed that average statistical power in neuroscience is very low, which inflates effect sizes and lowers reproducibility⁸. Add to that a pharmacological critique: very little of the oxytocin applied in the nose appears to reach the cerebrospinal fluid, while blood levels become supraphysiologic⁹.

The lesson for other peptides

Oxytocin is a useful example because it had what many trendy peptides lack: large, randomized trials published in leading journals¹³. When they arrived, the promised effects did not appear. Signals from small or animal studies for BPC-157, TB-500 or Semax deserve the same caution.