Autism spectrum disorder is a neurodevelopmental condition that affects communication, social interaction, sensory processing, behavior and daily functioning in different ways for different people. Autism is not caused by a single food, and no diet can be assumed to “cure” it.
Across multiple peer-reviewed human interventions, ketogenic and modified Atkins diets have produced measurable improvements in standardized autism assessments, including CARS, ADOS-2 and ATEC.[1–4] A newer randomized controlled preprint in children ages 2–5 adds further evidence of improvement when modified ketogenic nutrition is combined with rehabilitation therapy.[7]
Improvements have been reported in overall symptom scores, social affect, cognition, sociability, imitation and body use. The response is not uniform. The clinical question is therefore whether a carefully selected and monitored metabolic intervention can help a particular child’s symptoms, function or co-occurring medical conditions.
Why researchers study metabolism in autism
Ketogenic diets change the brain’s fuel supply, increasing ketone availability while reducing glucose and insulin exposure. Researchers are studying several possible pathways:
- Mitochondrial energy production
- The balance between excitatory and inhibitory signaling
- Oxidative stress and inflammatory signaling
- Gut-microbiome metabolites
- Brain-derived neurotrophic factor and related molecular pathways
- Seizure activity and abnormal cortical excitability
These mechanisms remain active areas of investigation. They provide biological reasons to study ketogenic therapy, but proposed mechanisms should not be confused with proof of clinical benefit.
The overlap between autism and epilepsy is particularly relevant. Some autistic children also have seizures or epileptiform EEG findings. Ketogenic diet therapy is already an established medical treatment for drug-resistant epilepsy, so a child with both conditions may have a separate neurological indication for dietary therapy.
Early human intervention data show improvements
The early intervention studies used different ketogenic formulations and different outcome measures, but each reported improvement in at least one standardized clinical assessment. The clearest signals involve overall CARS or ADOS scores, social affect, cognition and sociability.[1–4,7]
Modified Atkins diet compared with two other diets
The original Toward Health article cited a controlled study of 45 children ages 3–8 with autism spectrum disorder.[1] The children were divided equally among:
- A modified Atkins ketogenic diet
- A gluten-free, casein-free diet
- A balanced-nutrition control diet
After six months, both intervention groups improved on the Autism Treatment Evaluation Checklist and Childhood Autism Rating Scale compared with the control group. In the ketogenic group, average CARS score decreased by eight points, from 41.7 to 33.7. ATEC measures of speech, social function and cognition improved, and the ketogenic group showed greater improvement in cognition and sociability than the gluten-free, casein-free group.[1]
This study supports a clinically meaningful signal, but the groups were small—15 children each—and the dietary interventions were not identical except for carbohydrate restriction. The result supports further use and study of ketogenic therapy without implying that the same outcome will occur in every child.
Modified ketogenic, gluten-free diet with MCT
The second original citation reported an open-label intervention in 15 children ages 2–17 who completed three months of a modified ketogenic, gluten-free diet supplemented with medium-chain triglycerides.[2]
Autism Diagnostic Observation Schedule scores improved significantly after three months. Six children had a substantial improvement of more than 30% in total score, two had moderate improvement and seven had minor or no improvement. Average CARS-2 score decreased from 34.96 to 30.71. Significant improvements were also recorded for imitation, body use and fear or nervousness. Ten children assessed at six months maintained improvement in total ADOS-2 score and social affect.[2]
The study illustrates two important points: measurable improvement occurred, and response varied considerably among participants.
Additional human evidence
The original prospective pilot
In a 2003 prospective pilot, 30 children ages 4–10 began a six-month ketogenic intervention delivered in four-week diet periods separated by two-week breaks. Eighteen completed the protocol, and all 18 recorded some improvement on the Childhood Autism Rating Scale. Two improved by more than 12 points, eight improved by 8–12 points and eight improved by 2–8 points. In total, 10 children had improvements of at least eight CARS points.[3]
Seven children could not tolerate the diet and five discontinued early. This experience underscores both the potential for benefit and the practical importance of tolerability and professional support.
A 2026 intervention with behavioral and metabolic outcomes
A 2026 study enrolled 10 Indian children with autism and 10 age- and sex-matched neurotypical controls. The autistic participants completed a three-month gluten-free modified ketogenic intervention.[4]
Median Childhood Autism Rating Scale scores decreased by 4.5 points. The investigators also reported changes in plasma metabolites involving branched-chain amino-acid and one-carbon metabolic pathways, including decreases in leucine and coumarin and a modest increase in betaine. No diet-related adverse effects were reported during the study.[4]
This study adds current human evidence and connects behavioral change with measurable metabolic change. It was small and did not include a randomized dietary control group, so it is best understood as an additional clinical and mechanistic signal.
New randomized evidence in young children
A 2025 randomized controlled preprint enrolled 62 children ages 2–5. Both groups received rehabilitation therapy. The intervention group also received a modified ketogenic diet, while the control group continued a standard diet. Fifty-one children completed the two-month study: 24 in the ketogenic group and 27 in the control group.[7]
Autism Behavior Checklist scores decreased from 68 to 56 in the ketogenic group and from 67 to 58 in the control group. Childhood Autism Rating Scale scores decreased from 36 to 32 in the ketogenic group and from 35 to 33 in the control group. The between-group results favored the ketogenic intervention for both ABC scores (P = 0.015) and CARS scores (P = 0.010).[7]
Two children in the ketogenic group discontinued because of adverse reactions—one case of diarrhea and one rash—and three discontinued because of difficulty following the dietary protocol. Because this study is currently available as a preprint rather than a peer-reviewed publication, it is presented as emerging randomized evidence.[7]
What mechanistic follow-up studies add
Later analyses have examined biological changes within children receiving modified ketogenic interventions. These reports are valuable, but some involve the same or overlapping participants as earlier clinical work and should not be counted as wholly independent efficacy trials.
A 2020 metabolomic study evaluated 17 children with autism before and after a three-month modified ketogenic intervention. Ketones and acetylcarnitine increased, selenium levels changed toward those observed in neurotypical controls, and metabolic patterns differed between stronger and weaker behavioral responders.[5]
A 2024 analysis of seven children from a registered ketogenic intervention examined stool and blood before and after four months. Investigators reported reductions in two proinflammatory cytokines, changes in the gut microbiome and butyrate-related function, and changes in brain-derived-neurotrophic-factor-associated microRNAs.[6]
These findings help explain how ketogenic nutrition might influence brain and behavior through energy metabolism, inflammation and the gut-brain axis. They do not yet identify a reliable test that can predict who will respond.
What a medically supervised approach should include
A therapeutic ketogenic diet for a child is different from casually removing bread or sugar. The plan should protect growth, nutrition, feeding skills and family functioning.
Clinical preparation may include:
- Pediatric and developmental history
- Review of seizure history, EEG findings and medications
- Growth and nutritional assessment
- Identification of feeding aversions or severe food selectivity
- Baseline laboratory testing when clinically appropriate
- A defined dietary formulation and protein target
- Vitamin, mineral, hydration and bowel-management planning
- Tracking of symptoms, sleep, gastrointestinal function and behavior
- A scheduled reassessment to decide whether the intervention is helping
The chosen outcomes should be specific. Families might track social engagement, communication, adaptive function, sleep, gastrointestinal symptoms, seizure frequency or a standardized clinical score. Vague impressions make it harder to distinguish true response from normal variation.
What ketogenic therapy should not replace
Nutrition can be one part of a complete care plan. It should not automatically replace:
- Developmental and behavioral support
- Speech, occupational or educational services
- Treatment of epilepsy, sleep disorders, gastrointestinal disease or other co-occurring conditions
- Medication that remains helpful and medically indicated
Antiseizure medication should never be reduced without the prescribing clinician. Families should also avoid assuming that gluten, casein or carbohydrate restriction will address every symptom merely because those elements appeared together in a published intervention.
The takeaway
Human intervention studies show that ketogenic dietary therapy can improve selected autism-related measures in some children.[1–4,7] The strongest signals involve social affect, cognition, sociability and overall standardized scores.[1–4,7]
Response varies, and restrictive diets require careful nutritional planning. For families considering ketogenic therapy, the best approach is a defined, medically supervised trial with measurable goals and a clear plan for assessing benefit, tolerability and sustainability.
References
-
El-Rashidy O, El-Baz F, El-Gendy Y, Khalaf R, Reda D, Saad K. Ketogenic diet versus gluten-free casein-free diet in autistic children: A case-control study. Metab Brain Dis. 2017;32(6):1935–1941. doi: 10.1007/s11011-017-0088-z.
-
Lee RWY, Corley MJ, Pang A, et al. A modified ketogenic gluten-free diet with MCT improves behavior in children with autism spectrum disorder. Physiol Behav. 2018;188:205–211. doi: 10.1016/j.physbeh.2018.02.006.
-
Evangeliou A, Vlachonikolis I, Mihailidou H, et al. Application of a ketogenic diet in children with autistic behavior: Pilot study. J Child Neurol. 2003;18(2):113–118. doi: 10.1177/08830738030180020501.
-
Singh R, Shah A, Jain N, Shah H, Rawal R. Plasma metabolomic signatures in children with autism spectrum disorder and their modulation following a gluten-free modified ketogenic diet. BMC Psychiatry. 2026;26. doi: 10.1186/s12888-026-07917-1.
-
Mu C, Corley MJ, Lee RWY, et al. Metabolic Framework for the Improvement of Autism Spectrum Disorders by a Modified Ketogenic Diet: A Pilot Study. J Proteome Res. 2020;19(1):382–390. doi: 10.1021/acs.jproteome.9b00581.
-
Allan NP, Yamamoto BY, Kunihiro BP, et al. Ketogenic Diet Induced Shifts in the Gut Microbiome Associate with Changes to Inflammatory Cytokines and Brain-Related miRNAs in Children with Autism Spectrum Disorder. Nutrients. 2024;16(10):1401. doi: 10.3390/nu16101401.
-
Liu L, Zhou T, Ming L, et al. Therapeutic Efficacy of a Modified Ketogenic Diet in Autism Spectrum Disorder: A Randomized Controlled Trial. Research Square [preprint]. 2025. doi: 10.21203/rs.3.rs-6172026/v1.
