# Conduct disorder

How persistent conduct difficulties develop, what brain research does and does not establish, and age-appropriate treatment that supports safety and change.

Information, not medical advice. Discuss personal concerns with a qualified health professional.

[Watch the film](https://www.youtube.com/watch?v=XkgZCwF_MWQ) · [All films and essays](https://evidencepress.org/why-minds-break/films/)

Conduct disorder describes a persistent pattern of serious aggression, harm to other people or animals, destruction, deceit, theft or major breaches of rules appropriate to a young person's age. It concerns difficulties that substantially affect life, rather than ordinary disagreement or occasional misbehaviour. Understanding it requires attention to development, relationships and circumstances as well as the child's emotional and biological characteristics. There is no single cause shared by everyone with this diagnosis. [NICE guidance on conduct disorders](https://www.nice.org.uk/guidance/cg158/chapter/recommendations), [a developmental overview of conduct disorder](https://doi.org/10.1038/s41572-019-0095-y).

Research uses several overlapping descriptions: diagnosed conduct disorder, oppositional defiant disorder, aggression, antisocial behaviour and scores for broader conduct problems. These are not interchangeable. A study showing improvement on a parent's behaviour questionnaire does not necessarily show that a diagnosed disorder has ended, or that later offending has been prevented. This distinction matters throughout the evidence on causes and treatment. Most treatment evidence discussed here concerns children and adolescents under 18.

## Inherited susceptibility is real, but it is not destiny

Twin and family research suggests that inherited differences contribute substantially to conduct difficulties. Estimates often place heritability near 50%, although they vary with age, population and measurement. This describes differences between people within a studied population. It does not mean that half of one child's behaviour is genetic, that a child has a 50% chance of developing the disorder, or that half the problem cannot change. Inherited influences overlap with those involved in impulsivity, attention-deficit hyperactivity disorder, usually called ADHD, and substance problems. [Review of genetic influences on conduct disorder](https://doi.org/10.1016/j.neubiorev.2016.06.034).

Researchers have not found a single “conduct disorder gene”. Early studies often concentrated on individual candidate genes, including MAOA, and on whether particular variants altered the effects of adversity. A review of 87 studies of childhood aggression found an inconsistent picture. Aggression itself is also broader than diagnosed conduct disorder. These findings cannot support a genetic test that explains an individual child's behaviour or predicts their future. [Systematic review of childhood aggression genetics](https://doi.org/10.1038/s41398-024-02870-7).

Genes and surroundings do not operate in separate compartments. Some inherited characteristics influence the situations children encounter and the responses they receive. A child who finds it unusually difficult to pause, communicate frustration or tolerate disappointment may evoke different reactions from adults and peers. Adults' own characteristics also influence family life. This helps explain why an association between parenting and behaviour needs careful investigation; it does not make parenting irrelevant or excuse harmful treatment.

## Family experiences and social conditions can change development

Maltreatment, harsh or inconsistent discipline, family conflict, parental mental illness, poverty and exposure to violence are associated with conduct problems. None is necessary or sufficient to produce conduct disorder. Many exposed children do not develop it, and some children with the diagnosis have not experienced these circumstances. The same adversities also increase risks for other difficulties, including anxiety and depression. [Review of early conduct problems and their causes](https://doi.org/10.1111/jcpp.70031).

One important process is a repeating cycle in which conflict brings immediate relief to someone, making it more likely to happen again. A child may learn that escalating an argument makes a demand disappear. An exhausted adult may learn that shouting temporarily ends a confrontation. Each response can reinforce the other without either person choosing the long-term outcome. Communication difficulties, poor self-control, caregiver stress and lack of practical support can make these cycles harder to interrupt. Similar patterns can develop at school and among peers.

Research designed to separate inherited and environmental influences strengthens the case that circumstances matter, while leaving questions about direction. In a study of genetically identical twins, the twin experiencing harsher parenting also showed more antisocial behaviour. Shared genes could not account for that difference, but the study's measurements did not establish whether harsher treatment came first, difficult behaviour came first, or both affected each other. [Twin study of harsh parenting and antisocial behaviour](https://doi.org/10.1177/0956797620968532).

A natural experiment followed 1,420 children while some families received additional income. Conduct and oppositional symptoms improved among children whose families moved out of poverty. This is stronger evidence about environmental change than simply comparing richer and poorer families, although the key subgroup was much smaller than the full cohort, and the outcome was symptoms rather than recovery from diagnosed conduct disorder. A 2026 review of 48 longitudinal studies also supports connections between economic conditions and conduct problems, but finds that family stress and parenting do not explain every association consistently. Financial security, supportive relationships and access to resources can matter together. [Income and childhood mental health](https://doi.org/10.1001/jama.290.15.2023), [review of family economic circumstances and conduct problems](https://www.sciencedirect.com/science/article/abs/pii/S0272735826001212).

Another unusually informative study randomly assigned 136 young children living in institutions to foster care or care as usual. Among the 95 assessed later, boys assigned to high-quality foster care had fewer traits involving limited concern for others. This demonstrates that caregiving can affect development in circumstances of severe early deprivation. The finding was sex-specific, some children were lost to follow-up, and the setting was exceptional; it does not establish the same effect for every child or family. [Bucharest foster-care trial](https://doi.org/10.1016/j.jaac.2015.09.010).

Not every apparent environmental association survives stronger testing. Maternal smoking in pregnancy is associated with later conduct problems in conventional studies. Designs that better account for inherited and family influences provide much less support for smoking itself as an independent cause of those problems. That does not make smoking harmless: its effects on birth weight are well established. It shows why evidence about one outcome should not be transferred automatically to another. [Review of prenatal exposures and stronger causal research designs](https://doi.org/10.1017/S0954579418000421).

## The age of onset and emotional profile help describe needs

Difficulties beginning before age ten often follow a different course from difficulties first identified in adolescence. Early onset can signal more persistent developmental problems, but the distinction does not divide young people into two clean biological types. Later onset is not necessarily temporary or entirely caused by peers. Recalling exactly when symptoms began is also difficult, especially when early problems were missed. [Developmental overview](https://doi.org/10.1038/s41572-019-0095-y).

Some young people show persistently limited guilt, concern for others or emotional expression. Researchers often call these “callous–unemotional traits”; diagnostic descriptions use “limited prosocial emotions”. These terms describe a pattern requiring careful assessment across settings. They are not a verdict on a child's character or capacity to change. Most importantly, they do not describe everyone with conduct disorder.

Even within this group, emotional experience varies. Some children appear relatively untroubled by threat or distress, whereas others show considerable anxiety and histories of trauma. Aggression may be planned in pursuit of a goal, or may erupt during fear, frustration or perceived threat. These patterns can overlap. Boys and girls also share many risks while differing in some presentations and experiences. Such distinctions may guide assessment, but recent reviews caution against assuming that they reliably identify one specific brain process. [Review of conduct disorder and limited concern for others](https://doi.org/10.1056/NEJMra1315612), [review of sex differences in conduct disorder](https://doi.org/10.1111/jcpp.70234).

## Learning, emotion and the control of action

Several plausible processes help connect experience and biology to behaviour. A child who responds less strongly to another person's distress may receive a weaker emotional signal that an action is harmful. A child who quickly perceives hostility may react aggressively before checking what happened. Other difficulties involve learning from unwanted outcomes, stopping an action once started, or giving sufficient weight to consequences that will occur later. These are overlapping possibilities, not a single explanation for everyone.

A 2025 study compared 676 young people with conduct disorder and 742 without it. On a laboratory task, the conduct disorder group learned less from punishment and showed a greater tendency to initiate actions. Learning from rewards was not reduced. Here “punishment” means an unwanted outcome within a controlled learning task; it does not show that harsher discipline would help. Indeed, difficulty learning from an unwanted outcome gives no reason to assume that increasing its severity would solve the problem. [Study of action initiation and learning](https://doi.org/10.1016/j.biopsych.2025.03.005).

Brain research examines systems that support these functions. The amygdala helps evaluate emotionally important events; frontal brain networks contribute to planning and control; other connected regions help represent rewards and guide action. A 2023 review of functional brain scans found lower responses in the right amygdala during some tasks involving negative or fearful material. The result depended partly on the task and the way researchers analysed the brain. It does not mean that the amygdala is always underactive in conduct disorder, or establish that the difference caused the behaviour. [Review of emotion-processing brain scans](https://doi.org/10.1038/s41398-023-02363-z).

Larger structural studies find small average differences too. An international analysis compared 1,185 young people with conduct disorder and 1,253 without it. Differences in the surface area of parts of the brain were generally small, roughly 0.09 to 0.26 standard deviations. A standard deviation is a way of describing the spread of measurements: a difference of 0.26 does not mean a brain region is 26% smaller. There was substantial overlap between groups. Most findings remained after accounting for ADHD and intelligence scores, but a scan taken at one time cannot establish what came first. [International study of brain structure in conduct disorder](https://doi.org/10.1016/S2215-0366%2824%2900187-1).

A 2026 analysis of conduct-problem scores in 14,160 participants also found small relationships with brain structure, with statistical coefficients around −0.02 to −0.07. Results varied by sex and adjustment for other characteristics. Some 1,875 participants overlapped with the earlier analysis, so the two reports are not wholly independent confirmations. Together they support further study of development, rather than a brain scan capable of diagnosing an individual child. [International analysis of conduct-problem scores and brain structure](https://doi.org/10.1016/j.jaac.2026.03.002).

Variation within the diagnosis is especially important. In a study of 542 young people with conduct disorder and 710 comparison participants, 56% of the conduct disorder group showed no marked difficulty on three emotion-processing tasks. This does not demonstrate an absence of biological differences; three tasks cannot test everything. It does show why a group average should not be turned into an assumption about every child. [Study of different emotion-processing profiles](https://doi.org/10.3389/fpsyt.2020.585052).

## Stress responses do not provide a diagnostic test

Some studies associate lower resting heart rate with antisocial behaviour, but results depend on how behaviour and physiology are measured. Across 66 studies involving 10,227 participants, associations measured across a range of behaviour scores were not equivalent to consistently finding a low resting heart rate in diagnosed cases. During emotional tasks, findings also varied between physiological measures. The evidence does not support checking a child's pulse to explain their behaviour. [Review of heart rate and other physiological responses](https://doi.org/10.1016/j.neubiorev.2019.02.016).

Studies of cortisol, a hormone involved in the body's response to challenge, sometimes find a smaller response to an immediate social stressor among children with conduct or oppositional difficulties. This is different from having low cortisol all the time. It remains uncertain how much such differences precede the behaviour, follow repeated adversity or reflect adaptation to difficult circumstances. Research on testosterone, oxytocin and other signalling substances is mixed, and evidence about inflammation is sparse. There is no established biochemical signature shared by all children with conduct disorder. [Review of cortisol responses to social stress](https://doi.org/10.1016/j.neubiorev.2021.06.005), [review of signalling molecules and emotional traits](https://doi.org/10.1016/j.neubiorev.2025.106259).

## Treatment starts with relationships and everyday circumstances

The strongest treatment framework combines a careful assessment with structured help in the settings where difficulties occur. Assessment considers development, emotional experience, ADHD, autism, trauma and other mental health problems, substance use, language and learning, family relationships, school and social conditions. Helping with an accompanying difficulty can be essential to making the main intervention usable. Safety, cultural context, the young person's perspective and the needs of parents or carers all matter. [NICE assessment and treatment recommendations](https://www.nice.org.uk/guidance/cg158/chapter/recommendations).

Care also needs concrete goals, such as fewer aggressive incidents, safer relationships or better school participation. Progress and adverse effects should be reviewed against those goals. Access and competent delivery matter at least as much as a programme's label: a family needs help it can attend and use, followed by a clear next step if improvement is insufficient. Services considering a more expensive named programme need evidence that it adds benefit to the local care already available. [NICE quality standards for conduct-disorder services](https://www.nice.org.uk/guidance/qs59).

### Younger children: practical work with parents and carers

For children aged three to eleven, NICE recommends evidence-based parent or carer training, with individual work when group participation does not meet a family's needs. These programmes involve practice and feedback rather than simply telling adults to be stricter. They help carers give positive attention, notice and reward constructive behaviour, make instructions clear, establish routines and respond consistently without violence. Studies associate several of these components with better results, although comparing programmes cannot prove which ingredient caused the improvement. Offering this treatment does not establish that parents caused the disorder. [NICE recommendations](https://www.nice.org.uk/guidance/cg158/chapter/recommendations), [review of parenting-programme components](https://doi.org/10.1016/j.jaac.2018.07.900).

A 2026 review combined 64 trials in children and considered another 20 adolescent trials separately. On parents' reports immediately after treatment, parent-only programmes improved preschool behaviour by an average of about 0.61 standard deviations compared with waiting lists or usual care; the equivalent estimate for school-age children was 0.39. Programmes combining work with children and adults had estimates of 0.96 and 0.61 respectively. These suggest meaningful average improvements, with uncertainty around their exact size. They do not prove that adding a child component always makes a programme better: different programmes were tested in different samples, and longer-term findings were less consistent. [Meta-analysis of psychosocial interventions](https://doi.org/10.1542/peds.2025-072476).

Another review of 25 trials involving 2,023 children found improvements of about 0.64 standard deviations for parent management training and 1.22 for parent–child interaction therapy compared with waiting lists. The latter typically involves direct coaching during parent–child interaction. It was often studied in younger children, and only seven studies in the review used diagnosed disorders. These indirect comparisons cannot establish a universal winner. [Review of parent training and parent–child treatment](https://doi.org/10.1007/s10578-022-01367-y).

For children aged nine to fourteen, structured work on social understanding and problem-solving is another recommended option. Children aged three to eleven with severe, complex difficulties may need combined parent-and-child programmes. The evidence does not justify automatically adding the same child therapy for every family; direct comparisons are relatively small and needs differ. A 2026 pilot trial in children whose problems persisted after parent training is an example of efforts to develop next steps, but differing parent and teacher findings and the small sample prevent it from establishing a new best treatment. [NICE recommendations](https://www.nice.org.uk/guidance/cg158/chapter/recommendations), [pilot trial after unsuccessful parent training](https://doi.org/10.1371/journal.pone.0353611).

### Adolescents: coordinated help across settings

For young people aged eleven to seventeen, guidance recommends coordinated work involving the young person, family and wider environment, including school and peers. Multisystemic therapy and functional family therapy are organised approaches to this work. The principle of addressing several connected problems is well supported, but a named programme is not necessarily better than other competent services. [NICE recommendations](https://www.nice.org.uk/guidance/cg158/chapter/recommendations), [review of adolescent psychosocial treatments](https://doi.org/10.1080/15374416.2022.2145566).

A 2025 review included 16 multisystemic-therapy trials with 4,095 participants and seven functional-family-therapy trials with 1,471. Multisystemic therapy probably reduced time spent in care away from home, but did not show clinically meaningful advantages on most primary outcomes. Functional family therapy had some possible benefits, with less certain evidence. In England's START trial, involving 684 adolescents, five-year conviction rates were 55% after multisystemic therapy plus usual services and 53% after usual services alone. That difference was compatible with chance. It showed no superiority in that setting; it did not show that coordinated care is useless, because both groups received active help. [Review comparing multisystemic and family therapies](https://doi.org/10.1016/j.jaac.2024.10.008), [START five-year follow-up](https://doi.org/10.1016/S2215-0366%2820%2930131-0).

When a young person already requires placement outside the home, specialised treatment foster care can be an alternative to residential or group care. A review of eight controlled studies involving 633 young people found moderately certain evidence of fewer criminal outcomes and fewer days in locked settings. This is evidence about choosing between placements for particular young people, not a reason to remove children from safe homes. [Review of Treatment Foster Care Oregon](https://doi.org/10.1177/1049731519890394).

## Medicines have specific roles and need careful review

Medicines are not recommended routinely for conduct disorder itself. An important distinction is treatment of co-occurring ADHD. In children with ADHD and disruptive behaviour, stimulant medicines can also reduce aggression and opposition. One review found a parent-rated average benefit of about 0.55 standard deviations; teacher-rated effects were larger but more variable. Atomoxetine and guanfacine showed smaller effects on oppositional behaviour. These findings do not support giving ADHD medicines to children who do not have ADHD. [Review of ADHD medicines and disruptive behaviour](https://doi.org/10.1177/070674371506000202).

A stepped-treatment study illustrates why the sequence matters. Of 175 children enrolled, 151 completed an initial phase adjusting ADHD treatment alongside family work. Severe aggression settled in 96 children: 64% of those completing this phase, or 55% of everyone enrolled. Because this first phase lacked a control group, its improvement cannot be attributed to medication alone. Only 45 children entered the later randomised comparison of added risperidone, divalproex or placebo. Findings from that small selected group should not become a recommendation for routine extra medication. [Stepped treatment of ADHD and severe aggression](https://doi.org/10.1016/j.jaac.2019.12.009).

NICE reserves short-term risperidone for severe aggression with explosive anger and marked difficulty regulating emotions when psychosocial interventions have not worked sufficiently. It requires specialist assessment and monitoring of growth, weight, metabolic health, movement-related effects and prolactin, a hormone that the medicine can raise. Response should be reviewed after three to four weeks, with discontinuation if there is no clinically important response by six weeks. These are safeguards within clinical care, not instructions for families to alter medication themselves. [NICE risperidone recommendations](https://www.nice.org.uk/guidance/cg158/chapter/recommendations).

A Cochrane review of ten trials involving 896 young people found short-term evidence mainly for risperidone, with limited certainty and little information about lasting functioning. In two trials, children receiving risperidone gained about 2.37 kilograms more than comparison participants on average. Nine of the ten trials had industry support, which warrants attention to reporting and replication without automatically invalidating their findings. A 2026 review of three trials involving 279 children aged six to twelve with ADHD and disruptive disorders found that adding risperidone reduced aggression by about 0.79 standard deviations, while increasing prolactin. This narrow, short-term evidence does not establish long-term recovery or make risperidone suitable for all conduct problems. [Cochrane review of antipsychotic medicines](https://doi.org/10.1002/14651858.CD008559.pub3), [review of added risperidone in ADHD and disruptive disorders](https://doi.org/10.12809/eaap25137).

## Limited concern for others does not make a child untreatable

A review covering 60 studies and 9,405 participants found that children with more pronounced callous–unemotional traits improved in disruptive behaviour by a similar amount from before to after treatment as those with lower levels. They often began with more serious problems and still had greater difficulties afterwards. Because a before-and-after comparison is not the same as a controlled treatment comparison, this does not prove identical treatment effects in the two groups. It does challenge the claim that these children cannot improve. [Review of treatment and callous–unemotional traits](https://doi.org/10.1111/jcpp.13774).

Changing the traits themselves is a separate question. Controlled studies produced a small, uncertain average estimate of 0.09 standard deviations, with an uncertainty range from −0.02 to 0.20, compatible with no difference. Effective reduction of aggression should therefore not be confused with proof that every underlying emotional characteristic has changed. The appropriate response to persistent needs is further assessment and adapted support, rather than withholding treatment.

## Improvement can last, but some approaches cause harm

A review of 40 randomised parenting trials found that gains were generally maintained during follow-up, commonly over six to eighteen months. The estimated change from the end of treatment to follow-up was approximately zero: 0.01 standard deviations. That means little average deterioration or further improvement after treatment, not that treatment had no effect. Evidence over much longer periods, particularly for diagnosed conduct disorder and adult outcomes, remains thinner. [Review of the durability of parenting interventions](https://doi.org/10.1016/j.cpr.2016.11.006).

Intimidation can actively worsen outcomes. Trials of “Scared Straight” and similar prison-awareness visits found more subsequent offending, with an odds ratio of 1.68, rather than the hoped-for reduction. This describes higher odds, not 68 additional offenders per hundred young people. Participants were mainly boys at risk of, or already involved in, offending, rather than a pure conduct-disorder sample. The result concerns these intimidation programmes and cannot be used to evaluate every intervention called a boot camp, but it gives clear reason to exclude the studied approach from helpful treatment. [Cochrane review of Scared Straight programmes](https://doi.org/10.1002/14651858.CD002796.pub2).

## What the evidence can and cannot promise

The size of apparent treatment effects depends partly on who is studied and who reports the outcome. A more clinically focused review of 17 trials covering 19 interventions found smaller benefits: around 0.36 standard deviations in parent reports and 0.26 in teacher or independent ratings, with no clear self-reported benefit. Only three trials had assessments kept unaware of treatment allocation. Parents' observations are valuable, but parents participating in treatment also know what they received. School reports, young people's accounts and independent assessments add different information. [Review of psychological treatments for clinical conduct problems](https://doi.org/10.1111/jcpp.12590).

Other limits recur across the research. Comparisons with waiting lists often show larger differences than comparisons with good existing services. Randomising a whole treatment package does not identify its active ingredient. Improvement in aggression, school functioning, diagnosis and offending are different outcomes. Studies of mixed conduct problems may not apply equally to diagnosed conduct disorder. Treatment harms, family burden, dropouts and long-term physical effects are often less thoroughly measured than symptom change. Several reviews include the same original trials, so their results should not be counted as independent replications.

Research into causes faces related problems: early studies were often small and disproportionately involved boys; ADHD, trauma and substance use can influence findings; family reports and recalled onset dates are imperfect; laboratory tasks may not capture everyday life; and cross-sectional studies cannot separate causes from consequences. Larger samples help, but do not repair an unsuitable measure or untested causal assumption.

More decisive studies would follow children before problems begin, measure change repeatedly and test whether altering a proposed process actually improves life. Proposed biological tests would need to add useful prediction beyond a careful clinical assessment and work in diverse populations. Treatment research needs longer follow-up, realistic comparisons between services, and better evidence for girls, families with fewer resources and young people with ADHD, autism, trauma or persistent problems after initial treatment. Educational participation, relationships, safety, quality of life and family burden deserve attention alongside symptom scores.

A useful care plan does not require pretending that all the causes are known. It combines a careful developmental account with practical support, treatment matched to age and accompanying problems, and honest review of safety and progress. Conduct disorder names serious difficulties that deserve attention. It does not establish parental blame, an unchangeable brain defect or a fixed limit on a young person's future.

---

AI-assisted adaptation of supplied research; not independently clinically reviewed. Original essay text: CC0 1.0. Published 7 November 2026.