Muslim voters in Delhi are about 60% more likely than non-Muslims to be flagged for discrepancies in electoral records under India’s controversial Special Intensive Revision (SIR), an investigation by The Hindu has found.
The disparity is particularly pronounced in checks involving family size and age differences between relatives, raising questions about whether the Election Commission of India’s (ECI) automated verification system subjects Muslim voters to disproportionate scrutiny.
Muslims account for 13.6% of Delhi’s 97.54 lakh voters on the draft electoral rolls but make up 20.3% of those flagged for what the ECI calls “logical discrepancies”.
Their share rises to 30% among voters flagged under five criteria related to family size and age differences, according to the investigation.
Delhi’s draft electoral rolls contain 97.54 lakh voters. Of these, 33.12 lakh were flagged for notices during the revision exercise.
Another 13.79 lakh voters could not be linked to records from the previous revision, while 19.33 lakh were flagged for logical discrepancies.
Muslims accounted for just 7.6% of voters who could not be linked to earlier rolls. Their representation was considerably higher among those flagged for discrepancies.
For every 1,000 Muslim voters on the draft rolls, 318 were flagged for logical discrepancies. Among non-Muslims, the figure was 197, making Muslim voters roughly 60% more likely to be flagged.
The disparity widened in checks involving family relationships.
The ECI’s software flagged voters under several criteria, including cases in which six or more people identified the same person as a parent.
Other criteria covered parent-child age differences exceeding 50 years or falling below 15 years. Voters were also flagged when the age gap between grandparents and grandchildren was less than 40 years, or when siblings were born less than nine months apart.
Across these five family-size and age-related categories, 6.75 lakh voters were flagged. Muslims accounted for 30% of them, more than twice their share of Delhi’s electorate.
The sharpest disparity was found in cases where six or more people identified the same person as a parent. Of the 1.4 lakh voters flagged under this category, 58% were Muslims.
Across all five categories, 152 of every 1,000 Muslim voters were flagged, compared with just 56 among non-Muslims.
The analysis also found that nearly 60,000 voters had been flagged without any specific reason being recorded. Muslims accounted for almost 20% of this group.
The Hindu’s investigation found that the decisions to flag voters were made by ERONET, the ECI’s centralised software system, rather than by Electoral Registration Officers (EROs), who are legally responsible for maintaining electoral rolls.
A senior election official confirmed the software’s role to the newspaper on condition of anonymity.
The ECI has not publicly explained how the individual discrepancy criteria were developed or why particular voters were flagged. It has also not clarified whether being flagged automatically affects a voter’s eligibility.
The original SIR instructions for Bihar did not explicitly provide for issuing such notices to voters who had submitted enumeration forms and appeared on the draft rolls. The same was true of subsequent instructions for the second phase of the exercise.
However, logical-discrepancy checks were introduced in West Bengal during the second phase without a formal announcement, according to the investigation.
A directive issued on May 14, 2026, for the third phase referred to notices for voters whose records could not be linked to previous rolls or contained discrepancies.
The findings also raise questions about whether the family-size criteria adequately account for demographic differences between communities.
Data from India’s National Family Health Survey show that fertility rates have fallen substantially across religious groups. Among Muslim women, the total fertility rate declined from 4.41 children per woman in 1992–93 to 2.36 in 2019–22. Among Hindu women, it fell from 3.3 to 1.94 over the same period.
The gap between the two communities has narrowed considerably.
The Hindu’s findings also differ from the results of the earlier SIR exercise in Bihar, where Muslims accounted for 18.4% of voters deleted from electoral rolls, broadly in line with their 17.7% share of the state’s adult population in the 2011 Census.
Unlike the Delhi exercise, Bihar’s revision did not use the same logical-discrepancy checks.
For its Delhi investigation, The Hindu extracted voter information from draft electoral rolls using optical character recognition technology and matched the records against lists of voters flagged under different categories.
The newspaper estimated voters’ religion using a name-based statistical classifier tested across households in 17 Indian states. It also manually checked a random sample to assess the classifier’s accuracy.