Applied Behavior Analysis

See what research suggests this support may help with, how confident the evidence is, and where important limitations remain.

Behavior & LearningLast reviewed August 2026

What It Is

Understanding what the support actually involves.

Applied Behavior Analysis, or ABA, is a broad approach that applies principles of learning and behavior to understand why behaviors occur and to teach new skills or change specific behaviors. ABA can look very different across providers and programs. Approaches range from highly structured teaching such as discrete trial training and Early Intensive Behavioral Intervention to more naturalistic approaches. Some interventions combine behavioral principles with developmental approaches. Because ABA includes many different methods, evidence for one ABA based intervention should not automatically be assumed to apply to every form of ABA.

Quick Answer

What should you knowbefore digging into the details?

Applied Behavior Analysis, or ABA, uses principles of learning and behavior to teach skills and address specific behaviors. Research suggests that some ABA based interventions can improve language, adaptive skills, daily living skills, and targeted behavioral outcomes for some autistic children. However, ABA is not one standardized treatment, research quality varies substantially, outcomes are inconsistent, and important evidence gaps remain regarding harms, long term quality of life, adolescents, adults, and which individuals are most likely to benefit.

Best-Supported For

Language and some adaptive, daily living, and targeted behavior outcomes in young children

How Confident Are We?

Mostly Low certainty and highly dependent on the outcome and intervention

Biggest Limitation

Harms, long term quality of life, adolescents, and adults

Bottom Line

ABA based interventions can help some autistic children develop language, adaptive skills, daily living abilities, and specific functional behaviors. The evidence is strongest in young children and is considerably less certain for adolescents and adults. Confidence is limited by inconsistent intervention definitions, small or nonrandomized studies, risk of bias, differences in outcome measurement, and limited long term follow up. Research also does not adequately establish the frequency or magnitude of adverse effects. The goals, methods, intensity, provider quality, individual preferences, and respect for autonomy can substantially change what an ABA program actually involves.

Full Evidence

What does the research showfor each outcome?

Evidence can be stronger for one goal and weaker for another. Open any rating below to see why AFA reached that conclusion.

Outcome

Language and communication

Evidence certainty: LowFinding: Possible Benefit

Some ABA based early interventions improve language and communication outcomes in autistic children, but results differ across intervention types and studies.

Why did we rate it this way?

A recent systematic review and meta analysis found a moderate effect on language and rated the pooled evidence as moderate certainty.
That review combined Early Intensive Behavioral Intervention with naturalistic developmental behavioral interventions, so its estimate should not be assumed to represent every form of ABA.
The 2018 Cochrane review found improvements in expressive and receptive language but rated the evidence as low quality.
A 2020 meta analysis of randomized trials reported mixed findings across communication, expressive language, and receptive language outcomes.
Intervention methods, intensity, communication profiles, and outcome measures vary substantially across studies.

Outcome

Adaptive behavior and daily living skills

Evidence certainty: LowFinding: Possible Benefit

ABA based early interventions may improve adaptive functioning and daily living skills for some autistic children, although confidence in the size and consistency of the benefit remains limited.

Why did we rate it this way?

A recent mixed methods meta analysis found improvements in adaptive behavior and daily living skills compared with control conditions.
The review rated certainty for both adaptive behavior and daily living outcomes as low.
The 2018 Cochrane review also found improvement in adaptive behavior but rated the evidence as low quality.
A 2026 individual participant meta analysis reported meaningful adaptive behavior gains with Early Intensive Behavioral Intervention.
All studies in that 2026 analysis had serious risk of bias because participants were not randomly assigned.

Outcome

Social emotional functioning and behaviors of concern

Evidence certainty: LowFinding: Possible Benefit

Behavioral interventions can improve some targeted social emotional or behaviors of concern outcomes, but confidence decreases when studies with greater risk of measurement bias are excluded.

Why did we rate it this way?

The updated Project AIM meta analysis found a positive effect for behavioral interventions on social emotional or challenging behavior outcomes when analysis was restricted to randomized trials.
Much of that evidence depended on caregiver or teacher reported outcomes.
When Project AIM applied additional restrictions intended to reduce detection bias, evidence for behavioral intervention effects became less convincing.
Specific behavior analytic procedures may work well for narrowly defined behaviors even when evidence for broader developmental outcomes remains uncertain.
The reason a behavior occurs, communication access, medical factors, sensory needs, and environmental demands can affect whether a behavioral strategy is appropriate.

Outcome

Intellectual and developmental test scores

Evidence certainty: LowFinding: Possible Benefit

Some studies of intensive early behavioral intervention report improvements on standardized intellectual or developmental measures, but substantial methodological limitations make the size of this effect uncertain.

Why did we rate it this way?

The 2018 Cochrane review found improved IQ scores but rated the evidence as low quality.
A 2026 individual participant meta analysis found a substantial improvement in intellectual functioning among children receiving Early Intensive Behavioral Intervention.
Every study included in that analysis had serious risk of bias because treatment assignment was not randomized.
A 2020 meta analysis of randomized ABA based trials did not find a significant overall effect on IQ.
Changes in standardized developmental or IQ scores should not be assumed to represent equivalent changes in independence, well being, or quality of life.

Outcome

Measures of autism characteristics

Evidence certainty: Very LowFinding: Mixed Results

Research is inconsistent on whether ABA based interventions substantially change broad measures of autism characteristics, and reducing an autism severity score is not itself the same as improving quality of life.

Why did we rate it this way?

The 2018 Cochrane review did not find convincing evidence that Early Intensive Behavioral Intervention reduced autism symptom severity.
A 2020 randomized trial meta analysis did not find a significant overall effect on general autism symptoms.
A 2026 Early Intensive Behavioral Intervention meta analysis reported substantial reductions on autism severity measures.
The studies contributing to the 2026 estimate were nonrandomized and carried serious risk of bias.
Different studies define and measure autism characteristics differently.
The Atlas does not treat appearing less autistic as an automatic measure of meaningful benefit.

Outcome

Long term quality of life, mental health, and harms

Evidence certainty: InsufficientFinding: Effect Uncertain

There is not enough reliable evidence to determine the long term effects of ABA on quality of life, mental health, autonomy, or the frequency of adverse experiences.

Why did we rate it this way?

Long term quality of life and mental health outcomes are rarely included in controlled ABA effectiveness studies.
The updated Project AIM review found that adverse events across early autism intervention research were poorly monitored and reported.
Older EIBI reviews frequently reported no adverse events, but absence of reporting is not the same as evidence that adverse effects did not occur.
Qualitative and mixed methods studies include autistic people who describe both positive benefits and significant negative experiences.
Reports of trauma and psychological harm deserve attention, but current research cannot reliably estimate how frequently these outcomes occur or establish that every form of ABA carries the same risk.
Better prospective monitoring of well being, autonomy, stress, masking, adverse events, and quality of life is needed.

Who Was Actually Studied?

Applicability matters.

Age and life stage

The strongest effectiveness research involves young children, particularly preschool age children receiving early intervention. Evidence becomes much thinner for adolescents and adults.

Intellectual disability

Autistic children with developmental and intellectual disabilities are represented in parts of the ABA literature, but intellectual disability is not consistently reported or analyzed separately.

High support / profound autism

Children with substantial support needs are represented in early behavioral intervention research, but modern definitions of high support or profound autism are rarely applied consistently and subgroup outcomes are limited.

Minimal or no spoken language

Minimally speaking and nonspeaking autistic children are included in parts of the behavioral intervention literature, particularly communication focused research, but results are not consistently reported separately for this population.

Adults

Adults are poorly represented in controlled ABA effectiveness research. Adult research more often examines personal experiences, service history, attitudes, or specific behavioral supports rather than comprehensive childhood style ABA programs.

Settings

ABA based interventions have been studied in clinics, homes, schools, and community settings. Implementation, treatment intensity, staff qualifications, caregiver involvement, and opportunities for generalization vary widely.

What We Do Not Know

Important gaps remain.

Long term quality of life outcomes are rarely measured.
Autistic adolescents and adults are substantially underrepresented.
Adverse events and psychological harms are inadequately monitored in much of the intervention literature.
Research does not establish one universally appropriate number of ABA hours for every autistic person.
High support and profound autism are not consistently defined or analyzed separately.
Minimally speaking and nonspeaking participants are not consistently reported as distinct subgroups.
Many studies rely on small samples, nonrandomized designs, or outcomes vulnerable to observer bias.
ABA varies substantially between providers, making broad conclusions about all ABA difficult.
Autonomy, quality of life, stress, masking, and self reported well being are rarely primary outcomes.
There is limited evidence identifying which individuals benefit most from which specific behavioral strategies.

Potential Concerns

How a support is implemented matters.

ABA should not be treated as one standardized intervention. Goals, methods, provider quality, intensity, and implementation differ substantially.
Goals focused primarily on making an autistic person appear nonautistic may not represent meaningful improvements in communication, safety, independence, health, or quality of life.
Suppressing harmless self regulatory behavior solely because it appears unusual can conflict with the person's sensory or emotional needs.
Compliance should not be treated as the goal itself. Communication, refusal, boundaries, and the ability to say no are important skills.
The person's communication system should remain available and should not be withheld to force speech or compliance.
Programs should consider assent, autonomy, preferences, sensory needs, fatigue, and distress rather than relying only on observable behavior.
Very intensive schedules create substantial demands on the child and family, and research does not establish one appropriate intensity for every autistic person.
Provider training, supervision, ethics, and treatment quality vary, and recent stakeholder research suggests these differences affect people's experiences of ABA.
Because adverse effects are poorly monitored in the research literature, a lack of documented harm should not be interpreted as proof of safety.

Autistic Perspectives

Research evidence and lived experience answer different questions.

Published lived experience research describes highly varied experiences with ABA. Some autistic people, parents, and professionals report improvements in communication, independence, and useful skills. Others describe overstimulation, masking, loss of autonomy, coercive experiences, psychological distress, or trauma. These studies do not establish how common any one experience is, but they identify outcomes and ethical issues that effectiveness research has often failed to measure.

Themes Reported

Experiences differ substantially depending on the goals, methods, provider, intensity, and quality of the intervention.

Some participants report useful gains in communication, independence, and daily functioning.

Autistic adults in a 2025 mixed methods study were generally less satisfied with ABA derived interventions than parents and healthcare professionals.

Negative experiences include reports of overstimulation, masking autistic traits, excessive demands, psychological distress, and trauma.

Autistic participants emphasize the importance of being listened to and having communication, boundaries, preferences, and autonomy respected.

Higher perceived intervention quality was associated with greater satisfaction in recent stakeholder research.

Small qualitative studies can identify important experiences but cannot establish population wide rates of benefit or harm.

Lived experience is shown separately and is not mathematically combined with scientific evidence certainty ratings.

Sources Reviewed

See the research behind the review.

Individual Participant Data Meta Analysis2026

Clinically Significant Outcomes of Early Intensive Behavioral Intervention for Children With Autism Spectrum Disorders: An Individual Participant Data Meta Analysis

Seventeen studies were identified and participant level data were obtained from 15. The analysis found substantial effects for adaptive behavior, intellectual functioning, and autism severity, but all included studies had serious risk of bias because treatment assignment was not randomized.

View Source →

Systematic Review and Meta Analysis2026

Effectiveness and Experiences of Early Intensive Behavioral and Naturalistic Developmental Behavior Interventions for Autism Spectrum Disorders

Twenty five studies were included, including 16 randomized trials and nine qualitative studies. Benefits were reported for language, adaptive behavior, daily living skills, and joint attention. GRADE certainty ranged from moderate for language to low for several other outcomes. The review included multiple ABA based approaches, including EIBI and NDBI.

View Source →

Systematic Review and Meta Analysis2023

Autism Intervention Meta Analysis of Early Childhood Studies: Updated Systematic Review and Secondary Analysis

Project AIM included 252 controlled early intervention studies involving more than 13,000 participants. Behavioral interventions showed an effect on social emotional or challenging behavior outcomes in randomized trials, but many intervention effects weakened as analyses were restricted to studies with lower risk of bias. Adverse events were poorly monitored.

View Source →

Cochrane Systematic Review2018

Early Intensive Behavioral Intervention for Young Children With Autism Spectrum Disorders

Five studies involving 219 children were included. The review found possible improvements in adaptive behavior, IQ, and language, but rated the evidence low or very low quality because studies were small and mostly nonrandomized.

View Source →

Meta Analysis of Randomized Trials2020

Efficacy of Interventions Based on Applied Behavior Analysis for Autism Spectrum Disorder: A Meta Analysis

Fourteen randomized controlled trials involving 555 participants produced mixed results. Significant effects were not demonstrated for many broad outcomes, illustrating the variability of ABA based interventions and the limitations of drawing one conclusion across all outcomes.

View Source →

Mixed Methods Study2025

Experiences of Autistic Individuals, Caregivers and Healthcare Providers with ABA Derived Therapies: a Sequential Exploratory Mixed Methods Study

Focus groups and a survey of 219 participants found diverse experiences. Reported benefits included communication and independence, while concerns included overburdening, psychological effects, trauma, ethics, and quality control. Autistic adults were generally less satisfied than parents and professionals.

View Source →

Qualitative Lived Experience Study2023

Autistic Experiences of Applied Behavior Analysis

Seven autistic adults who received ABA in childhood described some benefits alongside significant negative experiences. The small qualitative sample cannot estimate population wide prevalence or establish the effects of all forms of ABA, but it identifies important lived experience concerns.

View Source →

Evidence Atlas

Evidence should help inform decisions, not make the decision for you.

Return to Evidence Atlas Read Our Methodology Contact AFA

Important: The Support Evidence Atlas is an educational decision support resource and does not provide individualized medical advice or prescribe a particular therapy, intervention, service, or support. Decisions should reflect the individual's needs, preferences, goals, abilities, environment, circumstances, and appropriate professional guidance when relevant.

Autism Foundation of America

What should you knowbefore digging into the details?

Important gaps remain.

Long term quality of life outcomes are rarely measured.

Autistic adolescents and adults are substantially underrepresented.

Adverse events and psychological harms are inadequately monitored in much of the intervention literature.

Research does not establish one universally appropriate number of ABA hours for every autistic person.

High support and profound autism are not consistently defined or analyzed separately.

Minimally speaking and nonspeaking participants are not consistently reported as distinct subgroups.

Many studies rely on small samples, nonrandomized designs, or outcomes vulnerable to observer bias.

ABA varies substantially between providers, making broad conclusions about all ABA difficult.

Autonomy, quality of life, stress, masking, and self reported well being are rarely primary outcomes.

There is limited evidence identifying which individuals benefit most from which specific behavioral strategies.

How a support is implemented matters.

ABA should not be treated as one standardized intervention. Goals, methods, provider quality, intensity, and implementation differ substantially.

Goals focused primarily on making an autistic person appear nonautistic may not represent meaningful improvements in communication, safety, independence, health, or quality of life.

Suppressing harmless self regulatory behavior solely because it appears unusual can conflict with the person's sensory or emotional needs.

Compliance should not be treated as the goal itself. Communication, refusal, boundaries, and the ability to say no are important skills.

The person's communication system should remain available and should not be withheld to force speech or compliance.

Programs should consider assent, autonomy, preferences, sensory needs, fatigue, and distress rather than relying only on observable behavior.

Very intensive schedules create substantial demands on the child and family, and research does not establish one appropriate intensity for every autistic person.

Provider training, supervision, ethics, and treatment quality vary, and recent stakeholder research suggests these differences affect people's experiences of ABA.

Because adverse effects are poorly monitored in the research literature, a lack of documented harm should not be interpreted as proof of safety.

Research evidence and lived experience answer different questions.