Ten Lakh Uniforms and Three Different Answers — India’s Internal Security Manpower Problem

Pritiraj BrahmaFounding Partner, Pritiraj & Partners · 19 August 2026

Parliament was told this year that the central armed police forces are short of 93,139 people, and also that they are short of 61,738. The same reply implies 69,246. A state that cannot state its own vacancy cannot plan its own recruitment.

ProvenancePublished 19 August 2026. CAPF strength figures are as placed before Parliament in 2026. State police ratios draw on successive volumes of BPR&D’s Data on Police Organisations, the most recent being as on 1 January 2024, released July 2025.

Three official answers were given to one question this year. In March, Parliament was told the central armed police forces and Assam Rifles were short 93,139 people. In July, it was told the shortfall was 61,738 — in a reply whose own sanctioned and actual strengths imply 69,246. None of the three has been reconciled.

Internal security is the part of national security that is delivered by staff rosters. There is no platform to induct, no trial to conduct, no foreign vendor to negotiate with. There is a sanctioned post, a person in it or not, and a deployment order. That makes it the cleanest possible test of whether the Indian state can execute an administrative task it fully controls — and the results are not good.

In 2026-27 the Ministry of Home Affairs has ₹1,73,803 crore for police, covering the central armed police forces, Delhi Police, Jammu and Kashmir Police, police infrastructure and the Intelligence Bureau (PRS Legislative Research, Demand for Grants 2026-27: Home Affairs). It is one of the largest single functional allocations in the Union Budget, and almost all of it is people.

Five findings anchor this analysis:

1. The vacancy figure is not stable across replies in the same year. In a Rajya Sabha reply in March 2026 the Minister of State for Home Affairs placed vacancies across the CAPFs and Assam Rifles at 93,139 — CRPF 27,400, CISF 28,342, BSF 14,531, ITBP 12,333, SSB 6,784, Assam Rifles 3,749. In a Rajya Sabha reply of 29 July 2026 the figure given was 61,738 against a sanctioned strength of 10,73,728, with actual strength of 10,04,482 — a difference of 69,246, which is neither of the other two numbers. The gap is almost certainly definitional, turning on which cadres and which categories of vacancy are counted. Nothing published says which.
2. The shortfall is concentrated in the two forces the country leans on most. On the July figures the CRPF stands at 3,12,707 against 3,30,841 sanctioned and the CISF at 1,74,567 against 2,00,548 — between them a little over 36,000 of the stated shortfall. The CRPF is the national reserve for internal disorder and the CISF guards critical infrastructure; these are the two rosters where a gap converts most directly into either unfilled duty or longer duty for those present.
3. Exit is rising even as recruitment continues. Resignations from the CAPFs rose from 1,255 in 2021 to 1,183 in 2022, 2,037 in 2023, 2,724 in 2024 and 2,333 in 2025, while voluntary retirements fell from 10,828 to 4,291 over the same period (figures placed before the Rajya Sabha in March 2026). That is not simply higher attrition; it is a change in its composition — fewer people leaving at the end of a career, more leaving in the middle of one. A force whose mid-career exits are climbing is telling its employer something about service conditions, and the reply does not engage with it.
4. Modernisation for these forces is a rounding error, and even that is underspent. Modernisation Plan-IV for the CAPFs, approved in March 2022, carried a total outlay of ₹1,523 crore to 31 March 2026 — against a police allocation of over ₹1.7 lakh crore in a single year. In 2022-23 the allocation was ₹248 crore and actual expenditure ₹78 crore; the Standing Committee on Home Affairs noted in 2023 that ₹22 crore had been spent by February, meaning ₹56 crore of the year’s spending occurred in its final weeks. The ministry’s stated reasons were procurement delays for technically complex items, tenders scrapped for not meeting requirements, and suppliers unable to deliver in time. Capital expenditure has held at 1 to 2 per cent of total CAPF spending for a decade (PRS, successive Demand for Grants analyses).
5. The upstream cause is the thinness of state police, which is a state subject nobody is required to fix. Successive volumes of BPR&D’s Data on Police Organisations put India’s police-population ratio at roughly 150 to 155 personnel per lakh, against the figure of about 222 per lakh commonly cited as a United Nations reference — a benchmark whose own provenance is looser than the frequency of its citation suggests. Vacancy across state police forces has stayed above 20 per cent for more than a decade on the same series. The Estimates Committee observed as early as 2018 that states depend heavily on central forces even for routine law and order. A Union force is being used to cover a state shortfall, and neither government is accountable for the other’s number.

The deployment loop

The mechanism is circular and each step is individually reasonable. A state’s police strength is below what its own sanctioned roster requires; when an event exceeds ordinary capacity — an election, a communal flashpoint, an agitation, a disaster — the state requisitions central forces. Central forces, deployed away from their own tasks, run their people harder. Harder-run people resign at higher rates in mid-career. Higher exits widen the very vacancy the deployment was covering for.

No participant in that loop is behaving badly. The loop persists because the cost of a requisition is not borne by the government that issues it, and the vacancy it creates downstream is reported by a different ministry, in a different reply, under a different definition.

Why the definition matters more than the number

It would be easy to treat 93,139 against 61,738 as a drafting inconsistency. It is worse than that. Recruitment at this scale runs on multi-year pipelines through the Union Public Service Commission, the Staff Selection Commission and the forces themselves; the size of the notification, the number of examination cycles and the medical and physical-test throughput all follow from an agreed vacancy figure. If the figure moves by 30,000 depending on the reply, then the plan built on it moves too, and no one can be held to a filling deadline because no one agreed what was to be filled.

The ministry’s stated remedies — radio-frequency identification during physical standard and efficiency tests, faster detailed medical examinations, regular departmental promotion committees — are sensible process fixes. They address throughput. They do not address the absence of a single authoritative number and a date by which it goes to zero.

The counter-case, honestly stated

The strongest defence is proportion. A shortfall of 61,738 against a sanctioned 10,73,728 is under 6 per cent — a fill rate most large public employers in India would envy, and far better than the 20-plus per cent that state police forces have carried for a decade. Absolute strength has grown substantially: the same forces stood near 10.05 lakh sanctioned in 2022 and are sanctioned at 10.74 lakh now, which means posts are being created as well as filled. Recruiting, testing and medically clearing tens of thousands of armed constabulary a year is genuinely hard, and the ministry’s process reforms are the right kind of unglamorous fix.

That case holds on the aggregate and fails on the concentration. A 6 per cent national gap that sits disproportionately in the CRPF and CISF, in forces that are also absorbing state-level deployment demand, is not experienced as 6 per cent by the battalion covering it.

What we would do

1. Publish one vacancy definition and one quarterly number. A single MHA statement of sanctioned, held and vacant strength by force and by cadre, with the counting rules stated on the page, refreshed quarterly. Where a reply departs from it, require the reply to say why. The reform is the definition, not the dashboard.
2. Attach a filling deadline with a named owner. For each force, a published target date to reach a stated fill rate, the officer accountable for it, and a quarterly variance report against the recruitment calendar. Successive assurances that recruitment is a continuous process have coexisted with a shortfall in the same range for four years.
3. Ring-fence a capital floor for the CAPFs and publish its utilisation monthly. One to two per cent of expenditure on capital, with the final weeks of the financial year carrying the bulk of it, is not a modernisation programme. A stated minimum share, procured on framework contracts rather than annual tenders, with monthly utilisation published, would end the March scramble that the Standing Committee has already identified.
4. Price the requisition. Publish central-force deployment days by requisitioning state, and charge states for non-emergency deployment at a published rate. A state that must pay for a battalion it uses in place of its own recruitment will recruit. Central assistance for election and disaster duty should remain free; ordinary law and order should not.
5. Tie state police strength to something that bites. Publish state-wise sanctioned-versus-filled strength alongside police-population ratio annually, and make a portion of Union modernisation assistance conditional on closing the state’s own vacancy rather than on submitting a plan to close it. Police is a state subject; the consequence of leaving it unstaffed is not confined to the state.

India’s internal security is delivered by roughly ten lakh people in central uniform and roughly twenty lakh more in state uniform. The country can describe its aircraft carriers with more precision than it can describe how many of those posts are filled. That is not a resourcing failure. It is a failure to agree on a number, publish it, and put a date beside it — which is the cheapest reform in this Review and, on the evidence of three answers in one year, still undone.

Companion essay: The Sanction Is Not the Squadron — the same architecture approached through a different instrument.