Six Crore Names: Inside India’s Largest Voter Roll Purge

On a Tuesday morning in October 2025, a 73-year-old man named Nabijan Mondal walked into his polling booth in Nandigram, West Bengal, and was told his name was not on the list. He had voted there for fifty years. The booth officer checked again, shrugged, and waved him away. Mondal stood outside for a long time before going home.

He was not alone. Across 14 states and union territories, nearly six crore names vanished from India’s electoral rolls between June 2025 and June 2026 — the largest single voter roll purge in the country’s history.

The exercise, called the Special Intensive Revision, or SIR, was launched by the Election Commission of India on June 24, 2025, beginning in Bihar. It spread to nine more states and three union territories in Phase II. Assam’s concurrent special revision brings the total coverage to roughly 14. Phase III, covering 16 more states including Maharashtra, Punjab, and Delhi, is underway now.

The Supreme Court upheld the exercise on May 27, 2026, in a unanimous 124-page judgment. The bench, led by Chief Justice Surya Kant and Justice Joymalya Bagchi, ruled that the ECI had the constitutional authority under Article 324 and Section 21(3) of the Representation of the People Act to conduct such a revision. The court applied a four-part proportionality test and found all four limbs satisfied.

But behind the legal architecture, a different story was unfolding — one of numbers that do not add up, of communities disproportionately affected, of a former CEC calling the exercise “completely unnecessary,” and of a 2.79-crore gap that the Election Commission has never explained.

The Numbers: State by State

The scale is staggering. In Uttar Pradesh alone, approximately 2.05 crore names were removed — a 13.2% reduction in the electorate. West Bengal lost roughly 91 lakh names, according to SABAR Institute’s analysis — though the ECI’s own pre- and post-roll totals show a net reduction of about 84 lakh, a discrepancy the commission has not explained. Gujarat lost 68 lakh. Tamil Nadu lost 74 lakh. Bihar, where the exercise began, lost 47 lakh names in its first phase.

Uttar Pradesh	15.44 crore	13.39 crore	~2.05 crore	-13.2%
West Bengal	~7.69 crore	~6.78 crore	~91 lakh	-11.9%
Tamil Nadu	6.41 crore	5.67 crore	~74 lakh	-11.5%
Gujarat	5.08 crore	4.40 crore	~68 lakh	-14.5%
Bihar	7.89 crore	7.42 crore	~47 lakh	-6.0%
Madhya Pradesh	5.74 crore	5.39 crore	~34.25 lakh	-6.0%
Rajasthan	5.46 crore	5.15 crore	~31 lakh	-5.7%
Chhattisgarh	2.12 crore	1.87 crore	~25 lakh	-11.8%
Kerala	2.78 crore	2.69 crore	~9 lakh	-3.2%

But the most striking number came not from the ECI but from an independent investigation. SabrangIndia, working with the Vote for Democracy collective, conducted a state-by-state reconciliation in June 2026.

They found that across the 14 states and UTs covered in Phase I and II, the pre-SIR electorate was 61.38 crore. The ECI declared 5.29 crore deletions and 1.87 crore additions through Forms 6 and 6A. Mathematically, the final roll should have been 57.96 crore. The actually published final roll was 55.17 crore.

That is a gap of 2.79 crore names. Vanished. Unexplained. The ECI has not addressed this discrepancy in any press release, affidavit, or public statement.

What the Supreme Court Actually Said

The petitioners — the Association for Democratic Reforms, Yogendra Yadav, CPI(M) leader Manoj Jha, TMC MP Mahua Moitra, the PUCL, and others — argued that the SIR violated Articles 14, 19, 21, 325, and 326 of the Constitution. They said it placed an unreasonable burden on citizens to prove their own citizenship, that it was rushed, that it excluded the most marginalised, and that it was timed to influence elections.

The court disagreed. In its judgment, it held that the SIR was “firmly anchored in both constitutional principle and statutory design” and traceable to Section 21(3) of the RP Act read with Article 324. It found the ECI empowered to conduct “a limited enquiry into citizenship for the purpose of satisfying itself as to eligibility for inclusion in the electoral roll” — though it clarified that this inquiry “does not amount to a determination of citizenship in the strict sense.”

The court laid down one safeguard: where the ECI is not satisfied that a person meets the statutory conditions, it must refer the case to a Competent Authority under the Citizenship Act within four weeks. The authority must decide before the next elections.

The court also said that deletion from the rolls does not mean loss of citizenship. In theory, this is a meaningful distinction. In practice, as reports from Bihar and West Bengal have shown, deleted voters have lost ration cards, cash transfers, and bank accounts. The welfare state moves faster than the judiciary.

The ADM Jabalpur Comparison

The PUCL called the judgment the “ADM Jabalpur moment” for voting rights — a reference to the 1976 Supreme Court decision that upheld the suspension of habeas corpus during the Emergency, widely considered the court’s darkest hour.

“Just as the Supreme Court betrayed the promise of the Constitution in ADM Jabalpur,” the PUCL said in a press statement, “the Supreme Court in ADR v. ECI lays waste to one of the key goals of the Constitution.”

The comparison is deliberately provocative. But the PUCL’s argument has substance. The court affirmed plenary ECI powers without addressing the structural accountability deficit — the fact that after the Anoop Baranwal judgment (2023), which was meant to make the ECI more independent, Parliament legislatively overturned key provisions. The ECI today, critics argue, has power without oversight.

Anmol Jain, writing in Verfassungsblog, called the judgment “Power Without Oversight.” Tarunabh Khaitan’s framework suggests the ECI lacks both adequate independence (executive capture of appointments) and accountability (no operational oversight mechanisms). M. Mohsin Alam Bhat argues the court’s “framework of trust” jurisprudence generates judicial deference rather than real scrutiny.

Who Got Deleted

This is where the story moves from numbers to people.

In West Bengal, SABAR Institute, a Kolkata-based research organisation, analysed the deletions constituency by constituency. They found that 34% of the 9.1 million names deleted were Muslim — against a Muslim population share of 27% in the state. That is a 7-percentage-point overrepresentation.

In Nandigram, where Muslims make up roughly a quarter of the population, 95% of the deleted names were Muslim. In Bhabanipur, where Muslims are 20% of the population, 40% of the deleted voters were Muslim. Murshidabad saw 460,000 Muslim deletions. North 24 Parganas saw 330,000. Malda saw 240,000. Overall, approximately 3.11 million Muslim names were removed from West Bengal’s rolls.

The Wire’s data analysis identified two distinct patterns of exclusion. In the Muslim-dominated border belt, disproportionately high numbers of voters were routed through an “under adjudication” pipeline — a scrutiny framework that, as The Wire noted, “carries its own burden of suspicion.” Constituencies with higher Muslim populations showed a “strong positive relationship” with the share of voters pulled into this process. In the urban and industrial belt, deletions were driven by the ASDD category — absent, shifted, dead, duplicate.

The border districts tell their own story. North 24 Parganas lost 1.26 million names. Murshidabad lost 749,000. In Malda and Murshidabad, villages displaced by river erosion — the Ganga and Padma rivers have been shifting for decades — found that “natural disaster begins to function as a bureaucratic ground for exclusion.” People whose land was eaten by rivers now found their names eaten by a database.

Women were disproportionately affected. Yogendra Yadav told Al Jazeera that the SIR placed “excessive burden on female voters.” In a patrilocal society, women move to their husband’s home after marriage but must produce documents from their father’s home. Name changes after marriage create official discrepancies. “In the eyes of law, it looks like a crime or fraud,” Yadav said.

The stories accumulate. Senarul Haque, a CRPF jawan who served for 35 years, was deleted from the rolls. Sadre Alam, a 1999 Kargil War veteran, was told his name was no longer on the voter list. Muhammad Daud Ali, a former Army technician, met the same fate. Suprabuddha Sen, an 88-year-old man whose grandfather illustrated copies of the Indian Constitution, was deleted without explanation. Hindustan Times reported three suicides among people who had migrated from Bangladesh and were facing SIR scrutiny.

Parakala Prabhakar, an economist, told The Guardian: “After this is completed, it will create two classes of Indians: those who are allowed full participation in political society and the political process — and those who are shut out. This is about killing the citizenship of minorities.”

In a separate interview with The Wire in April 2026, he added: “It is a bloodless political genocide.”

What the ECI Said

The Election Commission’s case rests on several pillars. The last intensive revision was in 2003 — more than 20 years ago. Urbanisation and migration have made rolls unreliable. Deceased voters remain listed. Duplicate entries exist across constituencies. Illegal immigrants are on the rolls, particularly in border states.

CEC Gyanesh Kumar’s stated goal was a “pure electoral roll” — “no eligible voters excluded and no ineligible persons included.” The ECI argued in court that Section 21(3) allows revision “at any time, for reasons to be recorded, in such manner as it may think fit.”

The process involved house-to-house enumeration by Booth Level Officers. The 2003 electoral roll served as the baseline — anyone whose name was on that roll was presumed eligible unless rebutted. Anyone not on the 2003 roll had to produce documents from a list of 11 prescribed categories. EPIC cards, the voter ID cards issued by the ECI itself, were initially excluded as proof. Aadhaar was added only after an interim Supreme Court order in September 2025.

But the PUCL pointed out that the ECI never answered the most basic question: why now? “What is the data which indicates the scale of multiple and defective entries on electoral rolls?” The ECI, they said, relied on “bald and bare assertions” without specific evidence.

SY Quraishi, who served as Chief Election Commissioner from 2010 to 2012, was more direct. “The SIR is completely unnecessary, it is designed to harass,” he told The Guardian. “It took us 30 years to achieve 99% accuracy in the rolls. They expect to exceed this in three months. Why this frantic rush if the main objective is accuracy?”

On the algorithm used to flag “logical discrepancies” — reactivated mid-exercise after the ECI had told the Supreme Court the de-duplication software was “far too defective to be deployed” — Quraishi said: “If software is being used to delete voters on the basis of these minor discrepancies, then it is a weapon against citizens’ rights and not fit for purpose.”

2003 vs. 2025: What Changed

The Wire’s investigation in October 2025 laid out the differences precisely.

The 2003 revision took about six months. The 2025 SIR was compressed to three months. The 2003 revision was completed well before the next Bihar elections. The 2025 SIR was completed roughly one month before the November 2025 Bihar polls. In 2003, enumerators were told “it is not the enumerator’s job to determine citizenship.” In 2025, citizenship verification was central. In 2003, training was done in advance. In 2025, training was done during the enumeration period. In 2003, the EPIC card was sufficient proof. In 2025, the EPIC was explicitly excluded.

Yogendra Yadav’s Elector-Population ratio analysis is the most damning piece of evidence. Before the SIR, Bihar’s EP ratio — voters relative to population — was 97%, close to global best practice. After the SIR, it dropped to 88%. A 9-percentage-point drop in a single revision cycle. “Unprecedented in the history of any electoral roll revision in India,” Yadav said.

The Welfare Linkage

The Supreme Court was careful to say that deletion from the electoral roll does not mean loss of citizenship. But the state’s welfare machinery did not get the memo.

In Bihar and West Bengal, deleted voters lost ration cards, cash transfers, and in some cases bank accounts. Welfare databases are linked to electoral roll data more tightly than the court’s clarification anticipated. A person deleted from the voter roll on a Monday could find their PDS ration stopped by Wednesday.

This linkage has not been independently adjudicated. The Supreme Court’s judgment did not address it. No court has yet ruled on whether tying welfare benefits to electoral roll status violates the court’s own clarification — or any constitutional provision.

No Other Country Does This

India’s SIR is globally unique. No other major democracy combines a compressed three-month timeline, a 20-year-old legacy baseline, citizenship scrutiny embedded in the electoral process, algorithm-based flagging without established protocols, proximity to elections, and the linking of electoral deletion to welfare benefit removal.

In the United States, voter purges happen continuously but at vastly smaller scales. Ohio’s purge — the subject of the Supreme Court case Husted v. A. Philip Randolph Institute (2018) — removed hundreds of thousands of voters over several years for infrequent voting. India’s SIR removed roughly 100 times that number in a single exercise.

In the United Kingdom, the transition to Individual Electoral Registration in 2014 dropped roughly one million voters — but the transition was gradual, not compressed into months, and did not involve citizenship verification.

In Germany and France, registration is automatic — the government adds eligible voters to rolls based on residence data. Australia has compulsory enrolment. None of these countries have a mechanism comparable to SIR.

The Brennan Center’s research found that countries with automatic registration have the highest accuracy rates because errors are corrected continuously, not through mass purges. India’s approach is the opposite: neglect the rolls for 20 years, then conduct a single massive purge.

The Opposition’s Case

Mamata Banerjee called it “a constitutional crime” and “a crime against the people of India.” She filed her own writ petition in the Supreme Court in February 2026, challenging the West Bengal SIR.

Rahul Gandhi called it “vote chori” — vote theft. He accused the ECI of “getting help from inside” and alleged collusion between the CEC and the BJP.

TMC MP Sagarika Ghose told The Guardian: “This will go down as a scandal in the history of post-independence India. However poor you are, however helpless you are, you have that right to vote. But that has been snatched away.”

The INDIA bloc collectively called the SIR “vote theft by government” and disrupted Parliament over it. MPs were detained during a protest march from Parliament to ECI headquarters in August 2025.

Notably, Shashi Tharoor broke ranks. He told the Times of India in September 2025 that “poll rolls are not perfect” and backed a nationwide SIR — a position his party did not endorse.

The Algorithm Nobody Asked For

The Reporters’ Collective revealed that the ECI reactivated its de-duplication and “logical discrepancy” flagging software mid-exercise in West Bengal — the same software the ECI had told the Supreme Court was “far too defective to be deployed.” No written protocols governed its use. No testing results were published.

In West Bengal, 2.7 million voters were placed in an “under adjudication” category — a liminal state where they could not vote in the 2026 elections but had not been formally deleted.

The Wire’s analysis showed that Muslim-heavy constituencies were disproportionately routed through this pipeline.

The Matua and Namashudra communities — Hindu refugees from Bangladesh who arrived post-2002 — found themselves trapped between SIR documentation requirements and a CAA backlog that has barely moved.

What Comes Next

Phase III is underway. It covers 16 more states and three union territories — Andhra Pradesh, Haryana, Jharkhand, Karnataka, Maharashtra, Punjab, Telangana, Uttarakhand, Delhi, and others — encompassing 36.73 crore electors. The data has not been published yet.

The SC’s four-week referral deadline for citizenship-doubtful cases has not been publicly reported as met. The number of cases referred to the Competent Authority, and how many have been adjudicated, is unknown.

The 2.79-crore gap remains unexplained. SabrangIndia has published its reconciliation. The ECI has not responded.

The BJP’s “Panna Pramukh” network — booth-level workers assigned to each page of the electoral roll — has been credibly alleged to have influenced enumeration outcomes in several states. This has not been independently scrutinised.

And the question that Nabijan Mondal, the 73-year-old from Nandigram, might ask if anyone bothered to listen: if he voted for fifty years from the same booth, in the same village, and his name was on the 2003 roll that the ECI itself declared as the baseline of probative evidence — how did his name disappear?

The Election Commission has not answered that question. Neither has the Supreme Court. The rolls have been published. The elections are being held. The names are gone.

Sources

• Election Commission of India press releases and PIB statements (June 2025–June 2026); Supreme Court judgment in Association for Democratic Reforms v. ECI, 2026 INSC 564 (May 27, 2026); SabrangIndia and Vote for Democracy state-by-state reconciliation (June 2026); The Hindu (multiple reports, June 2025–June 2026); The Guardian (April 22, 2026); Al Jazeera (April 16, 2026); BBC (April 2026); CNN (May 1, 2026); The Wire data analysis and dashboards (April 2026); PUCL press statement (June 2026); Verfassungsblog legal analysis (June 3, 2026); SC Observer case repository; Reporters’ Collective investigation; Hindustan Times (November 2025); India Today (November 2025); Progressive International (April 28, 2026); Brennan Center for Justice; MIT Election Lab; SY Quraishi interview in The Guardian.