Start with a single number.
According to the 2011 Census, about 2.21 per cent of India's population lives with a disability. Roughly 2.68 crore people.
Now place it beside another. The World Health Organization estimates that around 16 per cent of the global population, some 1.3 billion people, experience significant disability.
India is not an outlier in human biology. Nothing about the country's demographics, health profile or ageing curve explains a sevenfold divergence from global prevalence. Which leaves one plausible explanation, and it is the reason this article exists.
The gap is not in the population. The gap is in the counting.
That matters far beyond statistical tidiness. Budget allocations, the number of special educators a state trains, how many resource rooms a district plans, how many assistive devices are procured and how seriously disability is treated in policy conversations all follow from the number.
When the number is low, the response is small. And for intellectual disability in particular, the number is not just low. It is close to invisible.
A note on the figures below: every statistic is attributed to its source and year. India's disability data comes from instruments built at different times using different definitions, and they are not directly comparable with one another. Where figures are cited here, please verify against the primary source before reproducing them in official material.
What India's official numbers actually say
Three sources carry most of the weight, and understanding what each one is for is the beginning of reading any intellectual disability statistics India publishes.
The Census of India, 2011. The most comprehensive count, and the most dated. It recorded roughly 2.68 crore persons with disabilities, about 2.21 per cent of the population, with a majority in rural India and slightly more men than women. Crucially, it counted disability across eight categories: seeing, hearing, speech, movement, mental retardation, mental illness, any other, and multiple disabilities. The terminology alone tells you how old the framework is. The full tables are published by the Office of the Registrar General and Census Commissioner.
The National Sample Survey, 76th Round, 2018. A dedicated survey on persons with disabilities, conducted by the Ministry of Statistics and Programme Implementation. It arrived at a broadly similar overall prevalence of around 2.2 per cent, but it was the first major national instrument to work with the twenty-one conditions recognised under the Rights of Persons with Disabilities Act, 2016 rather than the old eight. It is a sample survey rather than a full enumeration, so it gives you rates and patterns, not headcounts by district.
The UDID database, ongoing. The Unique Disability ID system, run by the Department of Empowerment of Persons with Disabilities, issues certification-linked cards and publishes live dashboards broken down by state and by disability type on the UDID portal. This is the most current disability census India data substitute available, and for anyone writing, planning or fundraising in 2026, it is the number to check rather than the 2011 figure.
But UDID measures something different from a census. A census tries to count everyone. UDID counts everyone who applied for and received a certificate.
The distance between those two things is one of the most important gaps in this entire subject, and it is not random. It systematically excludes exactly the people least able to navigate an application process.
Intellectual disability: the category the data handles worst
Within an already undercounted population, one group is undercounted more severely than the rest.
The 2011 Census recorded approximately 15 lakh people under the category it called "mental retardation", the terminology of that instrument, not of current law. Against a national population of roughly 121 crore, that works out to about 0.12 per cent.
Now compare that with international literature. Global estimates of intellectual disability prevalence generally sit somewhere in the range of 1 to 3 per cent of the population, depending on the definition and threshold used.
Even taking the most conservative end of that range, one per cent of India's 2011 population would have been around 1.2 crore people.
The Census found roughly one in eight of them.
That is not a rounding error or a methodological quibble. It is the difference between a policy problem that is visible and one that is not. And it explains a great deal about why intellectual disability receives so little dedicated funding, so few trained professionals and so little space in public discussion relative to its actual scale.
There are structural reasons for the undercount, and they are worth spelling out because they are fixable.
What the numbers miss
The definition changed and the instrument did not. The RPwD Act 2016 expanded recognised disabilities from seven to twenty-one, bringing in autism spectrum disorder, specific learning disabilities, chronic neurological conditions, blood disorders including thalassemia and sickle cell disease, and more. India's last full enumeration used eight categories from a pre-2016 world. A child with dyslexia or an adult with sickle cell disease has legal standing under the Act and, in the most recent Census, no box to be counted in.
Stigma suppresses disclosure. Disability, and intellectual disability especially, is still concealed in a great many Indian households. A family worried about a daughter's marriage prospects, a son's employability or the neighbourhood's opinion does not volunteer the information to a stranger with a clipboard. Every survey of this kind depends on people being willing to say something they may have spent years not saying.
The question format defeats the subject. Census enumeration involves a short, standardised question asked at the doorstep, often of one household member answering on behalf of everyone. Physical disabilities that are visible get recorded reasonably well. Intellectual disability, psychosocial disability and specific learning disabilities frequently do not, because identifying them requires assessment rather than observation. A family that has never had a child formally assessed has nothing to report, even when they know something is different.
Undiagnosed is uncounted. This is the compounding factor. In districts without paediatric developmental screening, without accessible assessment services and without a specialist within reasonable travelling distance, intellectual disability simply never gets identified. A child struggling in Class 2 is called slow. She turns eighteen without a diagnosis, without a certificate, without a single entitlement claimed, and without ever appearing in any dataset. The absence of services produces the absence of data, and the absence of data then justifies the absence of services.
Certification data reflects access, not prevalence. A state with high UDID issuance may have more people with disabilities, or it may have better outreach, more certifying authorities, shorter queues and more effective awareness. Reading certification counts as prevalence gets this backwards. It rewards states that have made the process work and penalises those where families cannot reach a district hospital.
Nothing captures severity, support needs or outcomes. Even a perfectly accurate headcount would tell you how many people have a condition. It would not tell you how many are in school, how many completed schooling, how many are employed, how many receive therapy, how many have any support system at all when their parents die. Those are the questions that determine what a life looks like, and India has almost no systematic data on any of them.
Reading the state-wise picture without being misled
State-level figures are where intellectual disability statistics India produces get misused most often, so a word on how to read them.
Absolute numbers largely track population. The most populous states — Uttar Pradesh, Maharashtra, Bihar, West Bengal, Madhya Pradesh, Tamil Nadu — report the largest absolute counts, which tells you almost nothing except that more people live there. Any list of "states with the most disabled people" is, in practice, a list of big states.
Prevalence rates are more interesting and more easily misread. Reported prevalence varies noticeably across states, and the instinct is to treat a higher rate as a worse outcome. Often it is the opposite.
Higher reported prevalence frequently signals better identification: more screening, more certifying authorities, more awareness, less stigma, more families willing to say so.
A state reporting low disability prevalence may be a healthier state, or it may simply be one where a great many people have never been counted.
The honest way to use state-wise data is comparatively and cautiously. Ask what changed between two rounds in the same state rather than which state has the highest figure.
Check whether certification infrastructure expanded. Look at whether the categories being reported match current law.
Disability data Tamil Nadu planners actually need
Tamil Nadu, as one of India's more populous states, accounts for a substantial share of the national total in absolute terms, and the state maintains its own disability welfare administration alongside the national UDID system. For anyone working here, the current and useful disability data Tamil Nadu offers comes from the state's certification and welfare records and the UDID dashboards, not from the 2011 Census.
The more revealing question is not the state figure but the distribution beneath it. Chennai, Coimbatore and Madurai concentrate the state's specialists, therapy centres, special schools and assessment facilities.
A family in a block in Ramanathapuram, Perambalur or Sivagangai may be several hours from the nearest developmental paediatrician.
Identification follows services, which means the districts with the fewest services will always report the fewest cases, regardless of what is actually true of their populations.
That is the pattern to look for in any state-level dataset: the districts reporting the lowest numbers are frequently the ones with the least capacity to find them.
Number of disabled children in India: where the data thins out further
Child-specific data is the weakest layer of all, which is unfortunate, because it is the layer on which almost every useful intervention depends.
The 2011 Census did record disability by age group, and the commonly cited child figures come from those tables. But the same limitations apply with greater force.
Early childhood is precisely when intellectual disability, autism and developmental delay are hardest to identify without professional assessment and easiest for a family to hope a child will grow out of.
Any figure for the number of disabled children India reports from that Census should be read as a floor, not an estimate.
School data adds a second problem. Enrolment records tell you how many children with disabilities are on a register.
They do not tell you how many attend, how many receive any adaptation, how many have an individualised education plan or how many quietly stop coming in Class 6 and are never followed up. A child can be simultaneously counted as enrolled and entirely unserved.
And there is a hard edge at eighteen. Much of India's disability support architecture, including the right to free education for children with benchmark disabilities, is built around childhood.
What happens afterwards, guardianship, livelihood, long-term care, what occurs when ageing parents can no longer provide support is barely measured at all.
Families living that transition describe it as falling off a cliff, and there is very little data to describe it as anything, because nobody is systematically asking.
Why this is not an academic problem
It would be easy to treat all of this as a technical debate for statisticians. It is not, for four concrete reasons.
Budgets follow counts. Allocation for disability programmes, assistive devices, scholarships and pensions is argued for on the basis of how many people need them. A category recorded at 0.12 per cent of the population commands a proportionate share of attention.
Reservation depends on identification. The RPwD Act reserves four per cent of government vacancies for persons with benchmark disabilities, with one per cent set aside for a category that includes autism, intellectual disability, specific learning disability and mental illness. A reservation is only usable by people who hold certificates. Undercounting translates directly into unfilled reserved posts, which then get read as a lack of eligible candidates.
School planning depends on numbers. How many special educators a state trains, how many resource rooms a district builds, how much Braille and adapted material is procured — all of it is planned against projected demand. Underestimated demand produces underbuilt systems, and the children who then cannot be accommodated are treated as exceptions rather than as evidence of a planning failure.
Invisibility shapes attitudes. A condition that does not appear in national statistics does not appear in national conversation. Intellectual disability's absence from the data is part of why it remains absent from public debate, and that absence is self-reinforcing.
What better data would look like
The fixes are not mysterious.
A full enumeration built on the RPwD Act's twenty-one categories rather than the pre-2016 eight, so that the counting instrument matches the law.
Enumerator training that treats intellectual and psychosocial disability as things requiring careful questions rather than observation.
Question design that reduces the cost of disclosure, so families are not effectively asked to announce a stigma at their front door.
Routine developmental screening in the early years, delivered through existing anganwadi and primary health infrastructure, which would identify children and generate data as a by-product of doing something useful for them.
Certification made genuinely reachable, with camps and mobile assessment reaching blocks rather than requiring families to reach district headquarters.
And measurement that goes past headcount into outcomes: school completion, employment, therapy access, support arrangements after parents. Those are the numbers that would tell us whether any of this is working.
Where our work sits in this picture
At Paavai Foundation, our work with children with intellectual disabilities takes place in exactly the districts where the data is thinnest.
That is not incidental. The community development work runs across districts in Tamil Nadu including Tiruchirappalli, Madurai, Namakkal, Krishnagiri, Tiruvannamalai, Theni, Dindigul, Sivagangai, Karur, Tirupathur, Perambalur, Pudukkottai, Ramanathapuram and Virudhunagar, with presence extending into Telangana and Andhra Pradesh.
These are places where a specialist assessment can mean a day's travel and a day's lost wages, and where the children least likely to appear in a national dataset actually live.
Paavai ID Care, our dedicated initiative for children with intellectual disabilities, is built around lifelong rather than episodic support: safe and supportive living environments, special education and assisted learning, therapy, rehabilitation and functional support, continuous medical monitoring, emotional and psychological well-being, and the everyday life skills that build assisted independence.
We do not publish beneficiary figures we cannot substantiate, and we would rather describe our work accurately than quantify it impressively.
What we can say is that the argument of this article matches what the work looks like on the ground. Children who have never been assessed. Families who did not know a certificate existed, let alone what it unlocks. Entitlements written into national law and never claimed by anyone in the village. The practical shape of that reality is set out in Understanding the Needs of Children with Disabilities: A Guide to Care, Inclusion and Support.
If you are a parent, a teacher or a supporter, the single most useful thing you can do with this article is act on it. Start by understanding what the law already guarantees: read our complete guide to the RPwD Act 2016, covering every right, benefit and entitlement, including certification, reservation, free education and how to complain when a right is denied.
Every uncounted person is a person
Statistics articles usually end by calling for better data. That is the right conclusion here too, but it risks sounding bloodless, so let me put it differently.
The gap between 0.12 per cent and a realistic prevalence estimate is not a methodological discrepancy. There are roughly a crore of people who exist, who have intellectual disabilities, and who have not been counted by their own country.
Each of them has a household that made decisions without information, a school that was never resourced for them, a certificate they may never have obtained, entitlements they never claimed, and a future nobody planned for.
Somewhere in that number is a mother who does not know what happens to her son after she dies, because no system has ever registered that he is there.
Counting people is how a state decides they exist. Until India counts these people properly, everything built for them will be built to the wrong size.
FAQs
What are the latest intellectual disability statistics India has published?
The most comprehensive full enumeration remains the 2011 Census, which recorded approximately 15 lakh people under its "mental retardation" category, roughly 0.12 per cent of the population. The National Sample Survey 76th Round (2018) provides more recent sample-based prevalence estimates using the RPwD Act's twenty-one categories, and the UDID portal publishes current certification data by state and disability type.
How many people with disabilities are there in India?
The 2011 Census recorded about 2.68 crore persons with disabilities, roughly 2.21 per cent of the population. The WHO estimates global prevalence of significant disability at around 16 per cent, which suggests substantial undercounting in Indian data rather than genuinely lower prevalence.
Why is India's disability figure so much lower than global estimates?
Several reasons compound: the last full Census used eight disability categories rather than the twenty-one recognised under the RPwD Act 2016, stigma suppresses disclosure, doorstep survey questions cannot easily identify intellectual and psychosocial disability, and undiagnosed conditions in areas without assessment services are never recorded at all.
What is intellectual disability prevalence in India?
Census 2011 recorded about 0.12 per cent. International estimates generally place intellectual disability prevalence in the range of 1 to 3 per cent of the population, which indicates that Indian data captures only a fraction of the actual number.
Where can I find current disability data for Tamil Nadu?
Current certification-based figures are published through the UDID portal, which allows state-wise and disability-type filtering, alongside the state's own disability welfare administration records. The 2011 Census figures for Tamil Nadu are now substantially dated.
How many disabled children are there in India?
The 2011 Census recorded disability by age group, and those tables remain the most commonly cited source for child figures. However, early childhood is when intellectual disability and developmental delay are hardest to identify without professional assessment, so child figures should be treated as a floor rather than an accurate estimate.
Does a UDID card count as disability data?
UDID counts people who have applied for and received certification, not everyone who has a disability. It is the most current source available, but it measures access to the certification process as much as it measures prevalence.
Why does accurate disability data matter?
Budget allocations, the number of special educators trained, resource room planning, assistive device procurement and the usability of reserved posts under the RPwD Act all follow from official counts. Undercounting produces underbuilt systems.
