Somewhere in your screening reports sits the most comforting sentence in the hiring process: no criminal record found. Hiring managers read it as a fact about a person. It is not. It is a fact about a search: no matching record was located, in the sources examined, in the jurisdictions covered, under the identifiers used, on the date the search ran.
In most countries that gap between sentence and belief is modest. In India it is structural, because of a fact that surprises even experienced HR leaders: there is no national criminal records database that an employer can query. No central registry returns a citizen’s criminal history on request. What exists instead is an inference, assembled from court records scattered across thousands of courtrooms and a police verification system built for different purposes, and the quality of that inference depends entirely on how well the search behind it was designed.
This piece is an honest anatomy of that check: what it actually consults, where it structurally fails in both directions, and what a clean result genuinely establishes. It closes the series this newsletter has run on familiar checks whose ground rules changed, and it is fitting that it ends here, because this is the most familiar check of all, and the least understood.
There Is No Database. There Is a Federation of Courtrooms
Start with why the database doesn’t exist, because it is constitutional, not accidental. Police and public order are State subjects under India’s Seventh Schedule, so criminal records live where cases live: with state police forces and in the courts that hear them. The systems that do connect this landscape, the CCTNS network linking police stations and the Inter-operable Criminal Justice System joining police, courts, and prisons, are built for the justice system’s own use. Employers cannot query them. India likewise has no equivalent of the UK’s Disclosure and Barring Service, where the state issues criminal record certificates for employment, and even the national sex offender database created in 2018 is accessible to law enforcement alone.
What an employer’s criminal check actually consults, therefore, is the public exhaust of the justice system: court records. A check searches for the candidate’s name across district and taluka courts, high courts, the Supreme Court, and relevant tribunals, looking for cases in which they appear. Layered on top, sometimes, is police verification through the local station. That is the whole apparatus. It is worth comparing the three global models plainly: in the UK, the state discloses your record; in the US, screeners search public records built around case files; in India, screeners infer a record’s absence from litigation searches run on a name. The Indian model can be done well. But it is an inference, and inferences inherit every weakness of their inputs.
The Search Runs on Names, and Names Are Not Identifiers
The first weakness is the key the whole search turns on. Indian court records are searched primarily by name, qualified by father’s name and sometimes address, because no unique identifier links a person to their cases. There is no number that follows a citizen through the courts the way a UAN follows them through provident fund contributions, a contrast readers of this newsletter’s employment verification work will feel immediately.
Names fail as identifiers in two directions at once. They collide: the same name and father’s name recur across a district, let alone a country, which is how an innocent candidate ends up matched to a stranger’s case, an error that, unexamined, becomes a wrongful rejection. Court records frequently lack dates of birth, removing the easiest disambiguator. And names mutate: Indian names transliterate into English half a dozen ways, court clerks spell phonetically, initials expand and contract, and the candidate’s passport spelling may match none of them. Each variant unsearched is a corridor unlit.
So the identical method produces opposite failures. The false positive: a same-name stranger’s case attributed to your candidate. The false negative: your candidate’s actual case, filed under a spelling or in a district the search never touched, returning silence. Any screening programme that has run at scale in India has lived both, and the honest ones design against both.
Five Crore Cases, Unevenly Lit
Now the terrain itself. As of the government’s most recent figures, pendency across India’s courts has climbed to 5.39 crore cases, with the Supreme Court’s own backlog up 11 percent in a single year and nearly 4,900 judicial vacancies across the district judiciary. Parliamentary replies over the years have shown the composition: the large majority of district-court pendency is criminal, and more than a lakh cases have been pending for over three decades.
For screening, the backlog cuts two ways. It means the population of live, findable cases is enormous, a candidate with a pending matter is statistically unremarkable, and it means the records describing those cases are administered by a system running under chronic strain. Digitisation, through the eCourts project and its public face, the National Judicial Data Grid, has transformed searchability: NJDG now offers a single searchable window from taluka courts to the Supreme Court, and it is the reason a source-level criminal check is possible at all. But the light falls unevenly. Metro district courts are well digitised; older records and smaller districts lag; and whole categories of forum, consumer commissions, labour courts, company tribunals, sit outside the main grid entirely, each with its own records and its own gaps.
And one structural blind spot has nothing to do with digitisation. Court records begin where courts begin. An FIR under investigation, a case not yet before a magistrate, generates no court record to find. The gap between a police station’s paperwork and a court’s docket is real time, sometimes long time, and everything inside it is invisible to a court-record search, however well designed.
Police Verification Verifies Less Than It Reassures
Which raises the natural question: doesn’t police verification fill that gap? Partially, and less than its stamp suggests.
Employer-facing police verification is, in most states, a check against local station records tied to the address supplied, plus a character attestation, delivered on timelines that stretch from weeks to months and vary enormously by state. It sees what that station knows about that address. A case registered elsewhere, a residence not disclosed, a life lived across three cities, all sit outside its frame, which is precisely how a driver can carry a clean verification and a live case in a court two districts away. The Police Clearance Certificate issued through Passport Seva serves overseas purposes well, and matters greatly to India’s Gulf-bound workforce, but it is an instrument of its own system, not a general criminal history.
None of this makes police verification worthless. It makes it a layer: address-linked, locally deep, and narrow, valuable for residence-tied roles and mandated categories, misleading only when treated as the check rather than a part of one.
Finding a Hit Is Easier Than Reading One
Suppose the search works and returns a genuine match. The harder discipline begins there, because Indian litigation is not a moral X-ray, and a raw court result read reflexively harms employers and candidates alike.
Consider what the pendency actually contains. Cheque dishonour cases under the Negotiable Instruments Act number in the tens of lakhs, technically criminal, overwhelmingly commercial in substance. Matrimonial and family-adjacent matters, property and land disputes with criminal complaints riding alongside, cases where your candidate is the complainant rather than the accused: all surface in the same search. And above everything stands the distinction the system itself insists on: an accused person is not a convicted one, and with cases routinely pending for years, treating every live matter as guilt would disqualify a meaningful slice of the honest workforce.
A mature check therefore ends in adjudication, not discovery: what is the matter, at what stage, how old, in which role does the candidate appear, and what does it genuinely say about the job at hand. The finding is where judgement starts, not where it stops.
The Ground Is Shifting Under the Search
The encouraging part of this story is its direction. The eCourts project’s third phase, funded at Rs 7,210 crore, is pushing digitisation deeper into records and processes. The new criminal codes that took effect in 2024 mandate electronic registers, e-FIRs, and digital case records, slowly shrinking the paper gaps. NJDG has made the judiciary’s own data public enough to be recognised under the Ease of Doing Business initiative, and searchability improves every year.
What will not change soon is the structural constraint: the absence of a unique identifier binding a person to their cases. Until that exists, every Indian criminal check will remain a name-based search across a federation of records, and its reliability will remain a function of design: identifiers verified, variants covered, jurisdictions mapped to the life, sources current, matches confirmed, findings adjudicated.
A Clean Check Is a Well-Designed Search
Which lands us at the reframe this whole series has been building toward. Readers may remember the piece months ago on America’s Clean Slate laws, and its lesson that a clean US check now means “no legally reportable record located,” not “no history.” India’s version of that lesson is older and deeper: a clean check here has always meant “no match found by this search,” and the only question that matters is how good the search was.
That is not a counsel of despair. It is a specification. A criminal check whose identifiers were verified, whose name variants were covered, whose jurisdictions followed the candidate’s actual footprint, whose sources were searched at the court and logged by date, and whose matches were confirmed before they became findings, that check’s “clean” is worth relying on. Anything less is a sentence that sounds like certainty and describes a shortcut. The companion playbook builds the specification step by step, because in a country of 5.39 crore pending cases and no database, the search is the check, and the design is the diligence.
