Oxytocin and Autism: The Research Story

Oxytocin and Social Behavior

290 children

The largest randomized trial of intranasal oxytocin for autism, published in the New England Journal of Medicine, found no significant improvement in social functioning after 24 weeks of daily treatment.

Sikich et al., NEJM, 2021

Sikich et al., NEJM, 2021

View as image

Few peptides have generated as much hope in autism research as oxytocin. Between 2010 and 2020, dozens of small studies suggested that this nine-amino-acid neuropeptide could improve social cognition, increase eye contact, and reduce repetitive behaviors in people with autism spectrum disorder (ASD). The hypothesis was intuitive: oxytocin drives social bonding in mammals, people with autism sometimes have lower oxytocin levels, and spraying synthetic oxytocin into the nose appeared to boost social responsiveness in early experiments.^[1] Then, in 2021, the largest and most rigorous trial to date found no benefit at all.^[2] The story of oxytocin and autism is a case study in how promising early results can collapse under the weight of proper controls. It is also a story that is not over.

This article covers the full arc: the biological rationale, the early excitement, the trial failures, the subgroup analyses that still intrigue researchers, and where the field stands now. For deeper coverage of specific subtopics, see our articles on intranasal oxytocin for social anxiety, oxytocin and trust, oxytocin and depression, and oxytocin for PTSD. For a broader look at oxytocin biology, see Oxytocin: Far More Than the "Love Hormone".

Why Researchers Connected Oxytocin to Autism

The oxytocin hypothesis for autism did not emerge from nowhere. It built on three converging lines of evidence from the early 2000s.

First, oxytocin's role in social bonding was already well established in animal research. Prairie voles that received oxytocin formed pair bonds; those given an oxytocin receptor blocker did not. In 2005, Kosfeld and colleagues published a landmark study in Nature showing that intranasal oxytocin increased trust in a monetary investment game in healthy humans.^[8] That same year, Kirsch and colleagues showed that oxytocin reduced amygdala activation in response to threatening social stimuli, suggesting the peptide dampened fear responses during social encounters.^[9]

Second, brain imaging and behavioral studies linked oxytocin to specific social cognitive abilities that are impaired in autism. Domes and colleagues found in 2007 that a single dose of intranasal oxytocin improved performance on the Reading the Mind in the Eyes Test (a measure of emotion recognition from facial cues) in healthy men.^[10] This was exactly the kind of social skill deficit that characterizes autism.

Third, some studies reported lower oxytocin levels in blood samples from children with autism compared to neurotypical controls, though these findings were inconsistent across studies and the relationship between peripheral blood levels and brain oxytocin activity remains unclear.^[1]

These threads wove together into a compelling narrative: autism involves social cognition deficits, oxytocin supports social cognition, and people with autism may have less of it. The logical next step was to give them more.

The Early Positive Results (2010-2017)

The first clinical signals were genuinely encouraging.

In 2010, Andari and colleagues at the Centre de Neuroscience Cognitive in Lyon published a study in PNAS testing intranasal oxytocin in 13 adults with high-functioning ASD. During a simulated ball-tossing game where some virtual partners were more cooperative than others, participants on oxytocin showed stronger preference for the cooperative partner and reported greater trust. They also spent more time looking at the eye region of faces during a gaze task.^[3] The sample was tiny, but the results aligned perfectly with the oxytocin hypothesis.

Gordon and colleagues followed in 2013 with a study showing that a single intranasal dose of oxytocin enhanced activity in brain regions associated with social perception (including the medial prefrontal cortex and right temporoparietal junction) in 17 children with ASD during a social judgment task using fMRI.^[4] The brain was visibly responding to the peptide.

Other small studies found oxytocin increased eye contact during naturalistic social interaction^[12] and improved social interaction in young children aged 3-8 years at certain doses.^[13]

Then came the Parker 2017 study from Stanford, which offered a potential explanation for why results varied. In a trial of 32 children with ASD, Parker found that those with low pretreatment blood oxytocin levels showed significant improvement in social responsiveness on the Social Responsiveness Scale, while those with higher baseline levels did not.^[5] This suggested oxytocin might work only in a biological subgroup, not across the board.

The pattern across these early studies was consistent: small samples (13-40 participants), short treatment windows (single dose or a few days), controlled laboratory environments, and statistically significant but modest effects on social measures.

The First Cracks: Multi-Week Trials

As researchers moved from single-dose experiments to longer treatment regimens, the results became murkier.

Dadds and colleagues tested intranasal oxytocin in 38 boys with ASD aged 7-16 over 5 days. They found no improvement on parent-rated social behavior measures beyond what placebo produced. Caregivers who believed their children received oxytocin reported greater improvements regardless of actual group assignment, suggesting a strong expectancy effect.^[14]

Guastella and colleagues tested 12 weeks of oxytocin (24 IU twice daily) in 50 male youth with ASD and found no treatment-specific changes on the primary social cognition outcome (the Diagnostic Analysis of Non-verbal Accuracy). Secondary measures were similarly negative.^[15]

The Watanabe 2015 study in Brain was more nuanced. After six weeks of intranasal oxytocin (48 IU daily) in 20 adult men with ASD, the researchers found no significant improvement on clinical social symptom ratings. However, they observed changes in resting-state brain connectivity between the anterior cingulate cortex and dorsomedial prefrontal cortex that correlated with modest improvements in social reciprocity scores.^[16] The brain was changing, but clinical symptoms were not.

Bernaerts and colleagues tested four weeks of intranasal oxytocin (24 IU daily) in 40 adult men with ASD using a randomized, placebo-controlled design with a year-long follow-up. The primary outcome (social symptoms on the Social Responsiveness Scale) did not improve. However, exploratory analyses showed improvements in repetitive behaviors that persisted at the one-year follow-up, months after treatment ended.^[7] This was interesting because repetitive behaviors were not the primary target, and the durability of the effect was unusual.

A pattern was emerging. Single-dose studies in controlled settings showed effects. Multi-week trials measuring real-world social behavior generally did not, at least not on the primary outcomes they were designed to detect. For a deeper exploration of how oxytocin relates to anxiety in particular, see our article on intranasal oxytocin for social anxiety.

The SOARS-B Trial: The Definitive Negative

The Study of Oxytocin in Autism to Improve Reciprocal Social Behaviors (SOARS-B), published in the New England Journal of Medicine in October 2021, was designed to settle the question.^[2]

It was the largest and most rigorous trial of intranasal oxytocin for autism ever conducted. The numbers: 290 children and adolescents aged 3-17 with ASD, randomized 1:1 to oxytocin or placebo, 24 weeks of daily intranasal treatment, target dose of 48 IU/day. The primary outcome was the Aberrant Behavior Checklist modified Social Withdrawal subscale.

The result was unambiguous. After six months of daily treatment, there was no significant difference between oxytocin and placebo on the primary outcome. There was no significant difference on any of the secondary outcomes. Adverse event rates were similar between groups.

The trial addressed the major methodological criticisms of earlier studies. It was properly powered (earlier positive studies had 13-50 participants; SOARS-B had 290). It used adequate treatment duration (6 months versus days or weeks). It enrolled the age group most commonly studied (children and adolescents). And the dose (48 IU) was at the high end of what previous studies had used.

A 2021 NEJM editorial accompanying the study summarized the state of the field: the initial optimism was not supported by the weight of evidence.^[2]

Meta-Analyses: What Aggregated Data Shows

Three major meta-analyses have attempted to make sense of the contradictory individual trials.

Huang and colleagues published a multilevel meta-analysis in 2021 examining 30 studies. They found a small but statistically significant overall effect of oxytocin on social cognition outcomes (Hedges' g = 0.22), but no significant effect on social behavior measures in daily life. The effect was driven largely by single-dose laboratory studies; multi-dose clinical trials contributed little to the positive signal.^[17]

Kiani and colleagues published an updated systematic review and meta-analysis in 2023 focused on adults with ASD. Across 12 RCTs, they found no significant improvement in social function or repetitive behaviors with intranasal oxytocin treatment.^[18]

The most informative meta-analysis came in 2024 from Zhang and colleagues, who conducted a dose-response meta-analysis of 19 RCTs. Their analysis found that 24 IU/day was the only dose associated with statistically significant improvement in social outcomes (SMD = -0.33, 95% CI -0.56 to -0.09). Higher doses of 40-48 IU/day showed no benefit. For repetitive behaviors, no dose reached significance.^[6]

This dose-response finding is provocative. The SOARS-B trial used 48 IU/day and found nothing. If 24 IU is the optimal dose, the largest trial may have tested the wrong amount. However, this is a post-hoc finding from aggregated data, not a prospective test, and should be interpreted cautiously.

Why Single-Dose and Multi-Week Results Diverge

The most puzzling feature of the oxytocin-autism literature is the gap between acute laboratory effects and chronic clinical outcomes. Several explanations have been proposed, and they are not mutually exclusive.

Receptor desensitization. Chronic oxytocin exposure may downregulate oxytocin receptors, diminishing the response over time. Animal studies have shown that prolonged intranasal oxytocin can impair social behavior in some contexts rather than enhance it.^[1]

Measurement mismatch. Single-dose studies measure specific cognitive tasks (emotion recognition, gaze patterns, brain activation) in controlled settings. Multi-week trials measure parent-reported or clinician-rated behavior in daily life. These are different constructs. A peptide could sharpen emotion recognition in a laboratory task without translating into observable behavioral change at home or school.

Delivery uncertainty. How much oxytocin actually reaches the brain after intranasal administration varies between individuals and between studies. Differences in spray devices, head positioning, nasal mucosa condition, and breathing patterns all affect delivery. A 2013 review by Guastella and colleagues highlighted the lack of standardization in nasal oxytocin administration as a critical problem for the field.^[19]

Heterogeneity of autism. ASD is not a single biological condition. It is a behavioral diagnosis encompassing many different genetic and neurobiological profiles. Parker's 2017 finding that only children with low baseline oxytocin responded to treatment suggests the peptide might work in a specific biological subtype that gets diluted in broad clinical trials.^[5]

Context dependency. Oxytocin's effects on social behavior appear to depend heavily on the social environment. In a supportive, structured laboratory setting with clear social cues, oxytocin may enhance social processing. In the unstructured, unpredictable social world of daily life, that enhancement may not be enough to change behavior. Research on oxytocin and trust has similarly shown context-dependent effects.

Who Might Still Respond: Subgroup Signals

Despite the overall negative picture, several subgroup findings remain active areas of investigation.

Low baseline oxytocin levels. Parker's 2017 Stanford study remains the strongest evidence for biomarker-guided treatment. If replicated in larger trials, blood oxytocin levels could identify the subset of children with ASD who might benefit.^[5]

Young children. The Yatawara 2016 trial found that intranasal oxytocin improved caregiver-rated social interaction in children aged 3-8 with ASD, particularly at lower doses (12 IU twice weekly over 5 weeks).^[13] Early childhood may represent a window where the developing social brain is more responsive to oxytocin.

Specific symptoms rather than global function. Bernaerts' finding of durable improvements in repetitive behaviors but not social symptoms suggests oxytocin may affect specific neural circuits rather than producing a general social enhancement.^[7]

Lower doses. Zhang's 2024 dose-response meta-analysis finding that 24 IU outperformed 40-48 IU doses suggests an inverted U-shaped dose-response curve, where more is not better.^[6] This has not been tested prospectively.

A 2025 review by Ricchiuti and colleagues from KU Leuven argued that the field should shift from asking "does oxytocin work for autism?" to "for whom, at what dose, at what age, and in what context does oxytocin work?"^[20] This reframing acknowledges that the broad-stroke approach has failed while preserving the possibility of targeted benefit.

The Neurodivergent Critique

Recent scholarship has raised a different question entirely: even if oxytocin worked, should it be used this way?

A 2024 paper by researchers examining oxytocin through a neurodivergent perspective argued that the premise of oxytocin therapy for autism, increasing social behavior to match neurotypical norms, conflates difference with deficit. The paper noted that many autistic adults do not want their social behavior changed and that the outcome measures used in clinical trials (eye contact frequency, social withdrawal scores) reflect neurotypical expectations rather than autistic wellbeing.^[20]

This critique does not invalidate the biological research. Oxytocin does modulate social neural circuits, and understanding those circuits advances neuroscience regardless of clinical application. But it adds an important ethical dimension that the field is still grappling with.

For context on how neuropeptides relate to mood regulation more broadly, and why the relationship between peptide levels and complex behaviors is rarely straightforward, see our article on neuropeptides and depression.

Safety Profile

One consistent finding across trials is that intranasal oxytocin appears safe in the short and medium term. The SOARS-B trial found similar adverse event rates in oxytocin and placebo groups over 24 weeks.^[2] Tachibana and colleagues reported that long-term administration over several months was well-tolerated in early adolescent boys with ASD, with no serious adverse events.^[21] Anagnostou's 2012 trial in adults similarly found no concerning safety signals.^[22]

However, animal studies on chronic oxytocin exposure raise theoretical concerns about receptor desensitization and potential effects on pair bonding and social behavior with prolonged use.^[1] Long-term safety data beyond six months in humans is limited.

Oxytocin's peripheral effects include uterine contraction (it is used clinically to induce labor) and milk ejection. For more on these reproductive functions, see our article on oxytocin and breastfeeding. These peripheral effects have not been a significant safety concern in the intranasal doses used for autism research.

Where the Field Stands Now

The current state of oxytocin-autism research can be summarized in five statements, each supported by evidence:

1. Intranasal oxytocin does not produce broad, clinically meaningful improvement in social functioning across the autism spectrum. The SOARS-B trial and multiple meta-analyses are clear on this.

2. Single doses of oxytocin can modulate social cognitive processing in laboratory settings. This effect is real but has not translated to clinical benefit.

3. A biological subgroup defined by low baseline oxytocin levels may respond differently. This finding from Parker 2017 has not been replicated at scale.

4. Dose matters in ways researchers did not initially appreciate. The 2024 dose-response meta-analysis suggests lower doses (24 IU) may be more effective than the higher doses (48 IU) used in the largest trials.

5. The field is shifting from "one-size-fits-all" trials toward precision approaches. Biomarker stratification, age-targeted interventions, dose optimization, and combination with behavioral therapy are all being explored.

Research into oxytocin and social behavior extends well beyond autism. For related reading on how this peptide functions in social anxiety, trust and bonding, and trauma processing, see the linked cluster articles. And for a broader perspective on how neuropeptide Y plays a complementary role in stress resilience, see our article on that topic.

The Bottom Line

Sixteen years of research on oxytocin for autism produced early excitement from small studies, followed by failure in larger, more rigorous trials. The largest trial (290 children, 24 weeks) found no benefit. Meta-analyses confirm that any effect is small and inconsistent. The remaining open questions center on whether specific subgroups, defined by baseline oxytocin levels, age, or dose, might still benefit from a more targeted approach.

Frequently Asked Questions

Does oxytocin help with autism symptoms?

Based on current evidence, intranasal oxytocin does not produce clinically meaningful improvement in autism symptoms when given broadly. The largest randomized trial (Sikich et al., 2021, 290 children over 24 weeks) found no significant difference between oxytocin and placebo on social functioning measures. Some smaller studies have shown effects on specific laboratory tasks like emotion recognition, but these have not translated to real-world behavioral improvement.

Why was oxytocin thought to treat autism?

Oxytocin drives social bonding in mammals, and early research suggested people with autism sometimes have lower oxytocin levels. Studies from 2005-2010 showed that oxytocin increases trust (Kosfeld et al., Nature, 2005), improves emotion recognition (Domes et al., 2007), and reduces amygdala fear responses (Kirsch et al., 2005). These findings made it a logical candidate for a condition defined by social difficulties.

What was the biggest oxytocin autism clinical trial?

The SOARS-B trial, published in the New England Journal of Medicine in 2021, enrolled 290 children and adolescents aged 3-17 with autism. Participants received either intranasal oxytocin (48 IU/day) or placebo for 24 weeks. The trial found no significant benefit on any social or cognitive outcome measure, making it the definitive negative result in this field.

Can oxytocin improve eye contact in autism?

Single-dose studies have shown oxytocin can increase gaze toward the eye region of faces in people with autism in laboratory settings (Andari et al., PNAS, 2010; Auyeung et al., 2015). However, this acute effect has not translated into sustained increases in eye contact during daily social interactions in longer clinical trials. The gap between lab and real-world effects remains a central unresolved problem.

What is the right dose of oxytocin for autism?

A 2024 dose-response meta-analysis by Zhang and colleagues found that 24 IU/day was the only dose associated with significant improvement in social outcomes across 19 randomized trials. Higher doses of 40-48 IU/day, including the dose used in the largest trial, showed no benefit. This suggests an inverted U-shaped dose curve, but the finding has not been tested in a prospective trial designed to compare doses directly.

Is intranasal oxytocin safe for children with autism?

Across clinical trials lasting up to 24 weeks, intranasal oxytocin has shown a safety profile similar to placebo, with no significant increase in serious adverse events. The SOARS-B trial (290 children, 24 weeks) and the Tachibana 2013 study both reported good tolerability. Long-term safety data beyond six months is limited, and theoretical concerns about receptor desensitization from chronic exposure have been raised based on animal research.