Court records · 15 min read

How criminal court records work

A screening result is a statement about a search, not a statement about a person. This is what the search actually reaches, what it does not, and the dozen ordinary reasons a record can be sitting in a courthouse and still not appear.

A result is a statement about a search

The sentence a background check produces is almost always phrased as a fact about a person: no record found, or record found. Both are really statements about a search, and a search is a set of decisions about where to look, what to look for, and how hard to try. Change any of those decisions and the result changes, which is why two screening companies can hand you different answers about the same person on the same day and both be telling the truth.

That is not a reason to distrust screening. It is a reason to understand what you are buying, because the employer is the one who has to act on the result and the employer who is asked to explain the decision afterwards.

Three systems, not one

The first thing to get straight is that a criminal record is not one object stored in one place. There are three separate systems, maintained by three separate institutions, for three separate purposes. They overlap imperfectly and they contradict each other routinely.

The clerk of the court

The clerk of the court is the custodian of the case file. Every charge that becomes a case gets a docket number, and the docket records what was charged, what happened at each hearing, and how it ended. This is the authoritative record of a disposition, and when a screening company says it verified a record at the source, the source it means is almost always here.

The important limitation is in the definition of what the clerk holds. The clerk holds cases. An arrest that never became a case is not in the courthouse at all, and no amount of searching there will produce it.

The arresting agency

The police department or sheriff that made the arrest keeps its own records: the arrest report, the booking, the fingerprint card, the incident number. In a large city these are substantial files with their own numbering system that has nothing to do with the court's docket numbers.

This is where an arrest that never became a case lives, and it is a separate institution from the court. Whether you can obtain it, and under what conditions, is a matter of state public records law rather than anything about the courts.

The state repository

Most states run a central criminal history repository, usually under the state police or a dedicated bureau of identification. It exists for fingerprint based background checks, it is the state's counterpart to the FBI's national index, and it is what people mean when they refer to a “rap sheet”.

It is not a copy of either of the other two systems. It is an assembly, and it is only as complete as the reports that reached it.

Why the repository is thinner than it looks

Employers tend to treat a state criminal history check as the thorough option, and on paper it should be: it is fingerprint based, which means it matches on a biometric rather than a name, and that eliminates the whole category of error that name searches suffer from. The catch is what is in it.

Three things thin it out. Repositories are built from submissions, so a record only exists there if some other agency sent it. The submissions depend on fingerprints, so offenses that were resolved without a booking, by citation or summons, may never generate one. And state law varies on which offenses agencies are required to report at all, with minor offenses frequently exempt.

The result is measurable and it is larger than most people expect. A 2002 study conducted for the Defense Personnel Security Research Center compared what state repositories returned against what local agency checks found for the same people, and found the repositories surfaced a fraction of the offense information: about 43 per cent in California, 61 per cent in Florida, 41 per cent in Pennsylvania and 19 per cent in Indiana. The study is old and the systems have improved since, but the structure that produced those numbers has not changed: a repository can only report what reached it.

The same effect shows up in state audits. One Illinois review found a substantial share of arrest submissions did not match the state system exactly, and that overall completeness of the criminal history record fell well short of the whole. None of this is an argument against running the repository check. It is an argument against treating it as the only check.

How the courts themselves are organized

The second thing to get straight is that “the courts” are not one kind of institution either. State trial courts come in two tiers, and the distinction decides where a given record lives.

Courts of general jurisdiction hear the serious cases: felonies, and the larger civil claims. The names vary by state, and the variation is a trap in itself, because a state's “district court” may be its general jurisdiction court in one state and its limited jurisdiction court in another.

Courts of limited jurisdiction hear the smaller ones: misdemeanors, traffic, municipal ordinance violations, small claims, and the early stages of felony cases before they are bound over. These also go by a wide range of names, including municipal court, county court, justice of the peace court, magistrate court, and city court.

The numbers matter here. According to figures collected by the National Center for State Courts, there are roughly 15,500 state trial courts in the United States, of which about 13,500 are courts of limited jurisdiction and about 2,000 are courts of general jurisdiction. Limited jurisdiction courts outnumber general jurisdiction courts by roughly six to one, and about two-thirds of all state trial court filings arrive in them. State courts overall handle something close to 97 per cent of all judicial proceedings in the country, with the federal system taking the remainder.

Two qualifications on those figures, because court statistics are easy to misread. The counts move when a state unifies its courts, which reclassifies existing courts rather than creating or closing any, so a comparison across years can show a large change that nobody experienced. And the federal system runs in parallel throughout: 94 district courts, each with its own courthouse, its own clerk, and its own records, holding federal charges that will never appear in any state system.

The courthouse is not the jurisdiction

This is the distinction that causes more confusion than any other, and it is worth stating plainly: a county is where a court sits, not the territory over which it has authority.

Courts are organized by judicial district, circuit or county depending on the state, and those boundaries do not always follow county lines. A single judicial district may cover several counties. A large county may contain multiple courthouses. A city may run its own municipal court with a jurisdiction that is the city, not the county around it. And a state may have a separate set of district courts for a specific purpose.

So when a screening company advertises a county criminal search, what it is actually offering is a search of the courts that sit in that county. That is a reasonable proxy most of the time, because most cases are tried where the defendant lives and most defendants are arrested where they live. It breaks in the cases that matter:

• The arrest happened in a county the person was passing through, and the case was tried there rather than at home.
• The case was transferred on a change of venue or consolidated with a case from another county.
• The charge is one that is heard in a courthouse the county search does not reach, which is exactly the limited jurisdiction problem above.

This is also why federal searches are separate. A federal charge is filed in a federal district court and is not in the state system at any tier, so a complete search is not one search done well. It is a set of searches across systems that do not talk to each other.

How a name based search misses a record that is there

This is the part worth reading twice, because the failure modes are almost all mundane. None of them require anyone to have made a mistake in the courthouse, and most of them are consistent with a search that was performed competently and honestly.

Identity

• The name does not match. A maiden name, a name changed on marriage or divorce, a hyphen added or dropped, a middle name used on one document and not another. Court indexes are name indexes, so a name the applicant no longer uses is a name the index will not return.
• The name matches too many people. This one runs the other way and is worse when it happens: a common name returns other people's cases, and the employer acts on a record that belongs to someone else. This is the allegation at the center of a large share of screening litigation.
• Date of birth is missing or absent from the index. Older records are frequently indexed by name alone, so there is nothing to disambiguate with, and a cautious search that requires a date of birth match will drop records that a looser one would have caught.
• The name is misspelled in the courthouse system. The record exists, the person is the right person, and the index entry is wrong. No search on the correct spelling will find it.

Where the case went

• The wrong tier was searched. The misdemeanor is in a municipal court and the search covered the county's general jurisdiction court.
• The wrong county was searched. Arrest and trial happened in different places, and the search followed the residence rather than the case.
• It is a federal matter. Nothing in the state system will ever show it.
• It is a juvenile matter. Juvenile courts are a separate system, records are sealed or confidential by default in most states, and a standard adult criminal search will not reach them even where the charge would have been a felony for an adult.

How the case ended

• Diversion, deferred adjudication or a deferred sentence. The case ends in a dismissal on completion, and there may be no conviction to find even though the arrest and the charge are both real.
• Expunged or sealed. The record existed and was lawfully removed, which means a clean result is the correct result. Whether a given disposition is expungeable, and by whom, varies enormously by state.
• Dismissed, with the arrest still on the police record. The court file shows a dismissal. The arresting agency's record still shows the arrest. Which of those you get depends entirely on which of the three systems was searched.

The record itself

• It was never digitized. Courthouse indexes are built over time and unevenly. In some courts the online index covers recent decades, and anything older is a paper file the clerk has to retrieve by hand, or a ledger that only exists on site.
• The court is not online at all. Plenty of limited jurisdiction courts have no public index, which means a search of them is a records request or a courthouse visit rather than a query.
• The status is out of date. The record says a case is pending when it was resolved months ago, or shows a warrant that has been recalled. This is the specific problem §1681k addresses, by requiring the agency either to notify the consumer or to maintain strict procedures for keeping public record information complete and up to date.

What a clean result actually means

Putting those together gives a fair reading of a result that comes back clean. It means a search was run against a defined set of sources with defined criteria, and returned nothing. Whether that is reassuring depends entirely on how the set and the criteria were chosen, and those are the two things an employer can actually ask about.

It is not a statement that the person has no criminal history, and it should never be treated as a warranty. The honest version of the sentence is the one a good screening report writes on itself: what was searched, and what was not.

Where the law sits over all of this

Two statutory duties run alongside these mechanics and are worth knowing exist, though they are set out properly in the FCRA compliance article. §1681k requires an agency supplying public record information for employment purposes either to tell the consumer at the time, or to maintain strict procedures for completeness and currency. §1681d(d)(3) bars an agency from furnishing public record information about an arrest, indictment, conviction, civil judicial action, tax lien or outstanding judgment unless its accuracy was verified within the thirty days before the report.

That second rule is why the courthouse verification step exists as a distinct stage rather than as a slogan. It is also the reason to ask a provider what happens when a courthouse cannot be reached, since the statute states a requirement without saying what a vendor should do when the source will not answer.

There is a third layer on top, and it is not about accuracy at all. The EEOC takes the position that an arrest record on its own establishes only that someone was accused, and that reliance on conviction records carries a disparate impact risk that has to be justified by the nature of the job and the time elapsed. Several states bar arrest-only reporting outright. That layer belongs to fair chance law, and it applies to what you may do with a record rather than to whether you can find it.

What to ask a screening provider

The questions that separate a real search from a database lookup are not technical and do not require legal training to ask.

• Which jurisdictions and which court tiers do you search by default, and what does it cost to change that?
• When a name comes back, how do you decide it is the same person, and what happens when you cannot?
• What do you do about records older than your online index, or courts with no public index?
• Are dispositions verified at the courthouse, and how current is that verification when the report is issued?
• What does the report say about what was not searched?

A provider that answers those specifically is describing a process. One that answers with coverage counts is describing a database.

What we do

Our own position follows from the above. We treat a criminal search as a claim about scope, so our reports state the jurisdictions and tiers covered and the identifiers used, and a result we cannot match confidently comes back as an unresolved item rather than as a record. Where a court has no public index, we say so rather than reporting the absence as a clean result.

The screening program toolkit carries the policy and adjudication documents that put this into a process, including the individualized assessment worksheet for the cases where the record exists and the decision is genuinely a judgment about the person.

Sources

• National Center for State Courts, Court Statistics Project
• National Center for State Courts, State Court Organization
• Bureau of Justice Statistics, State Court Organization
• Administrative Office of the U.S. Courts, Court Role and Structure
• SEARCH, Report of the National Task Force on the Criminal Backgrounding of America
• Defense Personnel Security Research Center, state criminal history repository study (2002)
• 15 U.S.C. §1681e(b) (reasonable procedures to assure maximum possible accuracy)
• 15 U.S.C. §1681k (public record information for employment purposes)
• 15 U.S.C. §1681d(d)(3) (verification of public record information)
• 15 U.S.C. §1681c(a) (obsolete information)
• EEOC Enforcement Guidance No. 915.002, Consideration of Arrest and Conviction Records (2012)

Last reviewed 2026-09-27. Nothing here is legal advice; see the terms of use.