TL;DR: No trial has compared CBT-AR and FBT-ARFID directly. FBT-ARFID, in which parents run meals, has randomized-trial evidence in underweight children 6 to 12, and CBT-AR, in which the patient does the exposure work, has open-trial evidence from about age 10 into adulthood. The starting point depends on who can run meals, how medically urgent the case is and which mechanism maintains the restriction.
Two hypothetical referrals arrive in the same week. The first is a 7-year-old whose growth has slowed, who eats a short list of pale, dry foods and nothing else, and whose parents plate every meal and sit through every one. The second is a 19-year-old who has eaten the same small set of foods since early childhood, whose bloodwork has started to reflect it, and who lives in a dorm 200 miles from home. Both would meet criteria for avoidant/restrictive food intake disorder. The question a family or a patient brings to an evaluation is which of the two main protocols, CBT-AR or FBT-ARFID, fits, and no study has compared them directly.
The short answer
CBT-AR (cognitive-behavioral therapy for ARFID) puts the exposure work in the patient’s hands, while FBT-ARFID (family-based treatment for ARFID) has the clinician coach parents to run meals. For a child whose weight is low and whose caregivers can run meals, FBT-ARFID has the stronger trial evidence. For an adolescent or adult who can do exposure work in session and is not in acute medical danger, CBT-AR is the protocol built for that work. The condition itself is covered on the ARFID topic page.
CBT-AR at a glance
Jennifer Thomas and Kamryn Eddy developed CBT-AR at the Massachusetts General Hospital eating disorders program and published the manual in 2019 (Thomas & Eddy, Cognitive-Behavioral Therapy for Avoidant/Restrictive Food Intake Disorder). The manual describes a protocol for patients about age 10 and older, delivered in 20 to 30 sessions across four stages, in an individual format or a family-supported format for younger patients.
Stage one covers psychoeducation, self-monitoring and early changes to the eating pattern. Stage two is treatment planning: the patient and clinician choose the target foods. Stage three works on the maintaining mechanisms, with modules chosen to fit the patient: sensory sensitivity, fear of an aversive consequence such as choking or vomiting, and low appetite or interest in food. Stage four is relapse prevention. The exposure work differs by mechanism: a patient with sensory sensitivity tastes several new foods each session, a patient with fear of choking or vomiting works up a fear hierarchy, and a patient with low interest practices exposure to body sensations such as fullness. A fuller walk-through of the stages is in CBT-AR for adults.
FBT-ARFID at a glance
FBT-ARFID adapts family-based treatment, the approach to adolescent anorexia nervosa that James Lock and Daniel Le Grange manualized from the Maudsley family therapy, to ARFID. The central principle carries over: parents, coached by the clinician, take charge of the child’s meals for a period and of the gradual expansion of what the child eats, with the child taking on more responsibility as eating stabilizes and as age allows. In the published case description (Rosania & Lock, 2020, Frontiers in Psychiatry), the family brings a meal to an early session that includes foods the child always, sometimes and never eats, and the therapist coaches the parents through it; at home, parents present foods in small steps of increasing difficulty, and a second phase begins once the child is trying new foods consistently. The model was built with children and early adolescents, and neither of its two trials enrolled a child older than 12.
Comparing the two
| Factor | CBT-AR | FBT-ARFID |
|---|---|---|
| Who does the work | The patient, in individual sessions, with family support for younger patients | The parents, coached by the clinician |
| Age range studied | About 10 through adulthood | Children aged 5 to 12 (2019 feasibility study) and 6 to 12 (2026 trial) |
| Delivery | Individual; a family-supported format exists | Family sessions; parents run meals at home |
| Medical urgency | Fits when acute risk is lower | Fits when weight restoration is urgent and parents can run meals |
| Primary target | The maintaining mechanism, through exposure | Renourishment first, with parents in charge of meals |
| What it asks of the patient | More: the patient does the exposure | Less at the start: parents run meals in the first phase |
| Length | 20 to 30 sessions, four stages | 14 sessions over about four months in the 2026 trial |
| Evidence so far | Open trials in youth (2020) and adults (2021); no randomized trial published | A 2019 feasibility study of 28 families and a 2026 randomized trial of 98 underweight children aged 6 to 12 |
What decides: meals, urgency and the maintaining mechanism
FBT-ARFID assumes a caregiver who can be at the table for nearly every meal, over months. In clinical practice, whether a caregiver can run every meal often settles the starting protocol, at times more than the child’s age. A parent working two jobs, a child who moves between two households on alternating weeks, a sibling whose needs take a parent from the table: each shapes whether the model can be run as written.
In clinical practice, families who struggle with conflict, or who find it hard to say what they feel, can find FBT-ARFID harder to run, for a reason other than logistics. Running meals under the model takes sustained, skilled communication between caregivers and with the child, and a household where disagreement goes unspoken or turns into a fight will find that demand heavier than the schedule itself.
Age enters the decision through the same factor. In clinical practice, older teens often resist a plan in which parents run their meals, and that resistance moves a plan toward CBT-AR. FBT asks parents to take over eating decisions for a period to protect the young person’s health, a request that sits against the young person’s growing claim on their own choices, most sharply for an older adolescent. CBT-AR keeps more of the work in the patient’s hands, which suits a teen who wants it there, and because neither protocol resolves that tension, an evaluation weighs what a given teen will take part in: CBT-AR with the family in a supporting role, a family-supported arrangement the teen has agreed to, or parent-run meals for a defined period when the medical picture leaves little room.
Medical urgency also shapes the starting point: when a child’s weight is low and falling, nutrition is the first priority, and in the 2026 trial described below, underweight children aged 6 to 12 gained more weight with FBT-ARFID than with the comparison therapy. ARFID occurs at every body size, and many patients are not underweight; for them, urgency shows in bloodwork, in missed growth, or in a diet that has narrowed to the point where one discontinued product removes a meal. When the patient is medically stable enough for a slower, exposure-based process, CBT-AR has room to work on the mechanism rather than on weight.
A child whose restriction is mostly low appetite may do well with a parent-led structure that gets regular nutrition in. A patient whose restriction is fear of choking, or a particular sensory profile, often needs the targeted exposure and interoceptive work that CBT-AR specifies for each mechanism. How those presentations look after adolescence is covered in ARFID in adults.
When sequencing makes sense
A plan can use both protocols in sequence, with FBT-ARFID restoring weight until the patient is nourished enough for CBT-AR’s exposure work, or in parallel, with parents managing meals while the patient attends mechanism-focused sessions. The arrangement depends on how the patient responds and how urgent the medical picture is. The wider set of ARFID options, including medical and nutrition care, is laid out in ARFID treatment, and families in western Pennsylvania can read what this practice offers on the Pittsburgh ARFID treatment page.
What the evidence shows
As of November 2026, a search found no published randomized comparison of CBT-AR and FBT-ARFID. The two protocols have been studied separately, in different populations.
For CBT-AR, Thomas and colleagues published an open trial in adults in 2021 (Journal of Behavioral and Cognitive Therapy): 15 adults aged 18 to 55 started 20 to 30 sessions, 14 finished, ARFID severity fell, patients added new foods, and nearly half no longer met criteria at the end. An earlier open trial in patients aged 10 to 17 (Thomas et al., 2020, International Journal of Eating Disorders) reported the same pattern. Open trials have no comparison group, so they show that patients improved during the protocol and nothing about whether it outperforms another treatment.
For FBT-ARFID, the 2019 feasibility study by Lock, Sadeh-Sharvit and L’Insalata randomized 28 children aged 5 to 12 to FBT-ARFID or usual care and found a full trial workable, with moderate to large effect sizes on weight and severity favoring FBT-ARFID, though a feasibility sample of 28 cannot establish efficacy; gains in parents’ confidence that they could change their child’s eating tracked with symptom improvement. In 2026, Lock and colleagues published a randomized trial in the Journal of the American Academy of Child and Adolescent Psychiatry: 98 underweight children aged 6 to 12, assigned at random to telehealth FBT-ARFID or an individual psychoeducational motivational therapy, 14 sessions each. Both treatments reduced ARFID severity, and FBT-ARFID produced more weight gain. The comparator was not CBT-AR, every child was underweight, and none was older than 12, so the trial answers a narrower question than which protocol fits which patient.
A large naturalistic study from a virtual clinic (Hellner et al., 2025, International Journal of Eating Disorders) followed 783 patients, 532 youth and 251 adults, who received CBT-AR or FBT-ARFID. Symptoms improved in both age groups. Treatment was assigned by age and clinical need, and the authors state that the study did not compare the two protocols, so it cannot separate their effects.
Where a manualized protocol reaches its limit, most often when emotion regulation or family conflict keeps interrupting meals, adjunctive approaches are discussed in when FBT isn’t enough.
When to get urgent care
Some situations call for a physician before any protocol question. Fainting or near-fainting, chest pain, a very slow or irregular heartbeat, dizziness on standing, signs of dehydration, or a day or more without eating or drinking: contact a physician the same day. For chest pain or fainting, call 911 or go to an emergency room. In a mental-health emergency, call or text 988.
The 7-year-old’s parents will probably start at the table, and the 19-year-old, who shares the diagnosis, will probably start in an individual room with a list of foods chosen for exposure. The study that would say whether either was the right call has not been run.