Using the Social Security Death Index to Confirm a Death
One federal file, two legal products, and a shelf of commercial copies of unknown vintage — that is the whole subject in a line, and the name on the tin is the first thing to put right. There is no government website called the Social Security Death Index. The federal record is the Social Security Administration’s Death Master File, and what genealogy and people-search sites label the SSDI is a copy of an extract of it — taken at some point, by somebody, and not necessarily refreshed since. That single distinction explains why two sites disagree about the same person, why a death you know happened is not there, and why a record you did find may describe a file state from years ago.
The short version
The SSDI is a product name, not a public record. The federal record is the Death Master File, built from the Social Security Administration’s Numident, and the copy you search on a genealogy site is a derivative of an extract of it. Federal law splits that file in two. Since 42 U.S.C. 1306c took effect, deaths inside a three-calendar-year window are held back from anyone who has not been certified by the Department of Commerce; older deaths sit in a separate product anyone may buy. The record is thinner than you think. After November 1, 2011 the published layout carries a name, a Social Security number, a date of birth, a date of death and a one-character proof code — and nothing about place, because the residence and payment fields were struck out and left blank. Both directions of inference fail. A person absent from the index may simply never have been reported to SSA; a person present in it may not be the person you are looking for. A matching name and birth year is a lead worth chasing, not a confirmation, and the document that actually confirms a death is a state death certificate.
Watch: one federal file, two products, and a shelf of copies
A short walkthrough of where the data starts, what the law does to it on the way out, and why the version you can reach is rarely the current one.
Watch the walkthrough
What you are actually searching
Three layers sit between a death and the row you read on a screen, and each one can lose something.
Layer one is the Numident. The Social Security Administration keeps an electronic record for every person ever issued a Social Security number, and when a death report reaches the agency and matches its payment records, the death goes on that person’s Numident entry. The agency’s own Inspector General describes the mechanism in exactly those terms in its July 2023 audit, Numberholders Age 100 or Older Who Did Not Have Death Information on the Numident: SSA “matches death reports it receives from various sources against its payment records and inputs death information on the Numident.”
Layer two is the Death Master File. SSA builds it out of the Numident. There are two versions, and the audit is precise about the difference. The full DMF goes to federal benefit-paying agencies — the audit names the IRS, Veterans Affairs, Agriculture, Defense, the Centers for Medicare & Medicaid Services, the Office of Personnel Management, the Pension Benefit Guaranty Corporation, the Federal Retirement Thrift Investment Board and the Railroad Retirement Board — to help them catch fraud. A subset goes to the Department of Commerce, which sells it. Everything a member of the public can reach descends from that subset, and the subset is smaller than the full file by design.
Layer three is the commercial derivative, which is where the label “SSDI” comes from. Genealogy and people-search companies license the Commerce product, load it, index it, and put a search box on it. That box is a snapshot of a file state, and nothing obliges the company to keep it current or to tell you when it was taken. Two sites can disagree about the same person without either of them being broken, because they are showing you two different vintages of the same underlying file. If you have ever had a record appear on one platform and not another and assumed one was wrong, this is usually the whole explanation.
The statute defines the file more narrowly than you would guess
Congress wrote a definition. 42 U.S.C. 1306c(d) says that for the purposes of that section, “Death Master File” means “information on the name, social security account number, date of birth, and date of death of deceased individuals maintained by the Commissioner of Social Security, other than information that was provided to such Commissioner under section 405(r) of this title.”
Read that twice, because two things are hiding in it. First, the file is four data elements: a name, a Social Security number, a date of birth, a date of death. Not an address. Not a place of death. Not a cause. Everything else you have seen attached to an “SSDI record” came from somewhere other than this statutory definition. Second, the closing clause carves out everything SSA received under section 405(r) of Title 42 — which is the provision covering the death data SSA gets from the states under contract. State-reported deaths are excluded from the DMF by definition, not by policy choice that could quietly reverse.
That carve-out is the legal machinery behind the change everybody half-remembers. NTIS, the Commerce agency that distributes the file, published an SSA-authored notice about it, Change in Public Death Master File Records, which states that effective November 1, 2011 the data NTIS receives from SSA no longer contains protected state death records, because section 205(r) of the Social Security Act bars SSA from disclosing the state death records it receives through its contracts with the states except in limited circumstances. The same notice quantifies it: the historical public file held 89 million records, SSA removed roughly 4.2 million of them, and the file has since taken in about a million fewer records a year than it used to.
One file, two products, and a three-year wall
This is the distinction no consumer guide to the subject makes, and it answers the most common complaint about the index.
In December 2013 Congress attached a restriction to the Bipartisan Budget Act. Section 203 of that act, Public Law 113-67, is codified at 42 U.S.C. 1306c, and its opening subsection is short enough to quote whole: “The Secretary of Commerce shall not disclose to any person information contained on the Death Master File with respect to any deceased individual at any time during the 3-calendar-year period beginning on the date of the individual’s death, unless such person is certified under the program established under subsection (b).”
Commerce delegated the certification program to NTIS, and the rules live at 15 CFR Part 1110. The definitions section of that part, 1110.2, is where the practical answer sits, because it names the two products NTIS actually sells:
- Limited Access DMF — “the DMF product made available by NTIS which includes DMF with respect to any deceased individual at any time during the three-calendar-year period beginning on the date of the individual’s death.” This is the recent-deaths file. Certification required.
- Open Access DMF — “the DMF product made available by NTIS which does not include DMF with respect to any deceased individual at any time during the three-calendar-year period beginning on the date of the individual’s death.” This is everything older. Section 1110.100(b) says certification is not required to access it.
So the answer to “why can I not find a recent death” is usually not a bug and not a delay. It is that the product available to an uncertified buyer is defined by the exclusion of recent deaths. A death from eleven months ago is not late arriving; it is on the other side of a statutory wall until the three calendar years have run.
What certification costs and what it commits you to
Certification is not a login. Under 15 CFR 1110.102 an applicant must certify, under 18 U.S.C. 1001, that its access is appropriate because it has “a legitimate fraud prevention interest, or has a legitimate business purpose pursuant to a law, governmental rule, regulation, or fiduciary duty,” and must specify the basis. It must also have systems, facilities and procedures to safeguard the data, held to requirements reasonably similar to section 6103(p)(4) of the Internal Revenue Code — the standard that governs handling of federal tax return information. Under 1110.101(b) that safeguarding claim has to be attested in writing by an accredited third-party conformity assessment body, not self-declared.
The NTIS program page and its companion certification site publish the fees, and they are worth knowing because they explain the shape of the market. Every figure in this paragraph is a federal fee, charged by NTIS or by an accredited assessor operating under its rules, and payable by an organization applying for certification. NTIS charges $2,930 to process the subscriber certification form, renewable annually. The accredited third-party body charges $247 for the attestation NTIS requires, every three years. An organization whose assessment body is in-house must apply to NTIS for firewalled status, which is another $268 every three years. None of it is refundable, and none of it buys the data — the NTIS data subscription is billed separately.
Then there are teeth. Section 1110.200, tracking 42 U.S.C. 1306c(c), imposes a penalty of $1,000 for each improper disclosure or use, capped at $250,000 in a calendar year, with the cap lifted entirely where the conduct is willful — defined in the rule as “a voluntary, intentional violation of a known legal duty.” The penalty reaches downstream too: a person who receives Limited Access DMF data and misuses it is liable whether or not they are themselves certified. Section 1110.201 requires certified persons to submit to scheduled and unscheduled audits, in the field or at the desk, as a condition of keeping the certification.
Two consequences follow that are useful to anyone evaluating a vendor. NTIS maintains a public list of certified persons, so a company’s claim to hold current-death data is a checkable claim rather than a marketing line. And Congress closed the obvious workaround: 42 U.S.C. 1306c(e) says no federal agency can be compelled to disclose this information to an uncertified person, and directs that the section be treated as a withholding statute under the Freedom of Information Act’s (b)(3) exemption. A FOIA request is not a side door into the recent-death file.
What is in a record, byte by byte
NTIS publishes the fixed-width layout. It settles several arguments at once.
The Limited Access Death Master File Extract Output Record Specifications, dated 11/2011, describe a hundred-character fixed-width record. Here is the whole of it:
| Position | Field | Size |
|---|---|---|
| 1 | Blank, or A (add), C (change), D (delete) | 1 |
| 2–10 | Social Security number | 9 |
| 11–30 | Last name | 20 |
| 31–34 | Name suffix | 4 |
| 35–49 | First name | 15 |
| 50–64 | Middle name | 15 |
| 65 | V or P code (verified or proof) | 1 |
| 66–73 | Date of death | 8 |
| 74–81 | Date of birth | 8 |
| 82–83 | State / county code of residence | 2 |
| 84–88 | ZIP code, last residence | 5 |
| 89–93 | ZIP code, lump sum payment | 5 |
| 94–100 | Blanks | 7 |
The three struck rows are the point. A footnote on the specification says the layout was “revised November 1, 2011 to remove the State/County Code of Residence, Zip Code – Last Residence, and Zip code – Lump Sum Payment fields as a result of no longer publishing protected state records,” and a second footnote marks those positions “always blank after November 1, 2011.”
Why “last residence” was never a place of death anyway
Every consumer write-up of this subject still tells readers that the record shows “the last place the person was known to be living,” and treats that as a pointer to where the death occurred. It is wrong twice over. On any record from the current file those bytes are blank, so there is nothing to read. And on a pre-2011 record, where the fields do carry values, they never meant what people took them to mean. One field was a ZIP code associated with the last residence in SSA’s own administrative records — which is to say the address SSA had, which may be years stale, may be a child’s house, and may be a nursing facility three states from home. The other was the ZIP code to which a lump-sum death payment was sent, which is the address of whoever claimed the payment, not of the person who died. Neither field was ever a place of death, and a search built on the assumption that it was will look in the wrong county’s records. If where a death happened is genuinely the question you need answered, the medical and certificate route is a different discipline, and the walkthrough of how a cause and manner of death gets established sets that out.
The one character that tells you how good the report was
Position 65 carries a proof code, and the specification defines it. V means the report was verified with a family member or someone acting on the family’s behalf. P means a death certificate, or the required information from one, was observed. N or blank means either the record predates the use of V and P codes, or the source was described as fairly reliable but nothing meeting the V or P standard was produced.
Most consumer platforms do not surface that character at all, which means a record verified against an actual death certificate and a record entered on a plausible-sounding phone call look identical on the screen you are reading. And even a P is not a warrant. The specification’s own note to subscribers is unambiguous, and it is worth quoting exactly: “LIMITED ACCESS SUBSCRIBERS MUST NOT TAKE ANY ACTION AGAINST ANY INDIVIDUAL WITHOUT FURTHER INVESTIGATION TO VERIFY ANY DEATH LISTED, EVEN IF V OR P CODES ARE PRESENT. FURTHER, SSA DOES NOT GUARANTEE THE ACCURACY OF THE DEATH MASTER FILE.” The NTIS notice on the 2011 change says the same thing in plainer words: users “should always investigate and verify the death listed before taking any adverse action against any individual.”
D is for delete, and this is why a stale copy stays wrong
Position 1 carries an update indicator on the weekly and monthly files: A for an addition, C for a change, D for a delete. The specification describes D as covering “any SSN record previously on the DMF which was found to be erroneous and should be removed.” It also documents a second retraction mechanism — when a previously reported death is corrected, NTIS receives a record with zeros in the date of death and a delete indicator, and subscribers are told to process the delete and remove the death from their own copy.
Deletes are the buried reason a commercial index can be confidently, visibly wrong about a living person. Corrections are pushed as instructions in an update stream, and they only take effect if whoever holds the copy applies them. A licensee that loaded a full file once and never processed the updates keeps showing a death that the federal file itself withdrew. That is a different failure from a gap in coverage, and it points in the opposite direction: the record exists, it is displayed, and the government has already disowned it.
One more thing the pre-2011 layout tells you
The specification includes the code table that used to populate those struck positions, and the table is not only states. Alongside the fifty-plus state and territory codes — with New York City carrying its own code, 54, separate from New York State’s 33 — there is a block of country codes: Canada, Mexico, Europe, Asia, South America, Central America and the West Indies, Oceania, the Philippine Islands, American Samoa, Guam, the Northern Mariana Islands, and a catch-all for areas under United States administration. That tells you deaths outside the country did reach the file and were coded when SSA had a value. It also tells you the resolution was continental. A record coded 58 puts a person in Europe, not in a country, and after November 2011 it puts them nowhere at all.
What each source proves, and what it only suggests
Six things people offer as evidence of a death, sorted by what a careful reader can actually take from them.
| Source | What it genuinely establishes | What it cannot tell you |
|---|---|---|
| State death certificate | That a named person died, on a date, in a place, certified by an official registrarThe actual proof | Nothing much — but availability and who may order one vary by state |
| Limited Access DMF (certified subscribers) | That SSA holds a death report for that SSN, including deaths inside the three-year window | Whether the report was ever verified; accuracy is expressly not guaranteed |
| Open Access DMF (anyone) | The same, for deaths older than the three-calendar-year window | Anything about a recent death — recent deaths are excluded by definition |
| A genealogy site’s “SSDI” search | That a matching row existed in whatever extract that site loaded, whenever it loaded it | The vintage of the data, whether deletes were applied, or which fields are blank versus absent |
| Obituary or funeral notice | Identity, family relationships and a hometown — the details that let you match a person, not just a name | Whether one was published at all; a great many deaths never generate one |
| Probate filing | That a court has accepted a death for the purposes of administering an estate, with named heirs | Anything about the majority of deaths, which never reach probate |
The ordering of that table is the argument. The document that settles the question is the one issued by the registrar who recorded the death, and everything above it in a typical search is a way of finding out where to send that request. Used that way — as a pointer rather than a verdict — the index is genuinely valuable, because a date of death and a state narrow a certificate request from an impossible search to a specific one. Used as an answer, it produces confident mistakes. The broader question of establishing whether someone has died, across all the sources at once, is worked through in the guide to checking whether a person has died; this page is about the federal file underneath one of those sources.
Why a death that certainly happened is not there
The gaps are structural, they are documented by the agency’s own auditors, and they are larger than most users assume.
Begin with the sentence that ought to be printed above every search box built on this data. In the July 2023 audit, SSA officials told the Inspector General that the agency “collects death information to administer its programs, not as a comprehensive accounting of all deaths in the country.” That is the design intent, stated by the people who run it. The file exists so SSA can stop paying benefits to people who have died. Completeness as a national death register was never the objective, and nothing in the system’s incentives pushes toward it.
The audit then measures one consequence. At the time of the review, roughly 18.9 million numberholders born in 1920 or earlier had no death information on their Numident record and consequently, in the auditors’ words, “were not included in any version of the DMF.” SSA had issued approximately 531 million unique Social Security numbers as of March 2023, so the missing set was about 3.6 percent of all Numident records. Officials also noted that almost none of those 18.9 million were receiving payments. That last detail is the tell: if a person is not drawing benefits, an unreported death costs the agency nothing, and the machinery that would otherwise catch it has no reason to run.
The same audit records what happened when the problem was flagged earlier. In 2015 the Inspector General reported 6.5 million numberholders aged 112 or older with no death entry, at a time when only 35 people worldwide were known to be that old. SSA added death information to about 1.5 million of them, explored the feasibility of correcting the rest, and ultimately decided not to.
The gaps, stated as categories
- The death was never reported to SSA at all. Reports come from families, funeral homes, hospitals, financial institutions and the states. A person who was not receiving benefits, whose family had no reason to contact the agency, generates no report and therefore no record.
- The death was reported only by a state. Section 205(r) bars SSA from disclosing state-supplied death records, and the statutory definition of the file excludes them. Those deaths are known to SSA and invisible in every public product.
- The death is inside the three-year window. Not missing — withheld, by 42 U.S.C. 1306c, from anyone Commerce has not certified.
- The death is old enough to predate systematic electronic records. The file was assembled from an administrative system, and coverage thins the further back you go.
- The copy you are searching is not current. The derivative may be years behind the federal file, and it may never have processed the deletes and corrections the federal file issued.
- The name does not match the way you spelled it. The layout allows twenty characters of surname and fifteen of given name, so long names truncate, and a married name, a maiden name or an anglicized spelling can hide a record from a search of the wrong version.
Put together, those categories support one conclusion, and it is the conclusion the ranking pages on this subject reach for and then soften. Absence from the index is not evidence that a person is alive. It is evidence that no report meeting the file’s conditions reached the file. Those are very different statements, and treating the first as though it were the second is how a family concludes an estranged relative must still be out there, and how a creditor keeps pursuing a person who died four years ago.
And presence is not proof of identity
The harder half of the argument, and the half almost nobody writes down.
The absence problem gets discussed. The presence problem does not, and it produces worse outcomes, because a hit feels like an answer in a way that an empty result never does.
Think about what a match actually consists of. You searched a name and a birth year. A row came back with the same name and the same birth year. What you have established is that at some point a death report reached SSA for a person recorded under that name and that date of birth. You have not established that the person is yours. Common surnames paired with common given names recur in every cohort; the layout truncates long names, which manufactures collisions that did not exist in the source; a date of birth in SSA’s records is whatever was written on a form decades ago, and transposed digits in a year are among the most common errors in any administrative file. Meanwhile the fields that would let you discriminate between two candidates — a place, a residence, anything geographic — are the exact fields struck out in 2011.
There is one field left that looks like a discriminator, and it is the one to be most careful with. The record carries a Social Security number. If you already hold the right person’s number from a document you obtained lawfully, comparing it is a legitimate check. What you must not do is run the process backward: take a number out of a death index, attach it to a living person on the strength of a name match, and treat that as identification. That is how a living person acquires someone else’s death record in a private database, and it is a difficult thing to get undone.
The standard we hold ourselves to
A name and a birth year agreeing is a lead. A confirmation is at least two independent records that agree on something beyond the name — a death date matching an obituary that also names the surviving relatives you expect, a probate case opened in the county where the person lived, a cemetery or interment record with the same date, a benefit termination consistent with the date. Best of all, and dispositive, is a death certificate from the state registrar. Where those disagree, the disagreement is the finding and it goes in the report as one. We would rather hand back an unresolved file than a confident wrong answer, because the actions taken on the back of a death confirmation — closing an account, distributing an estate, ending a benefit, telling a family — are not easily walked back. Where a matter turns on identifying the people who inherit rather than on the death itself, the discipline for that is set out separately in the guide to confirming that a claimed beneficiary is the right person.
The record that actually confirms it
Brief, because it is a subject in its own right — but you should know the door is not always closed.
Deaths are registered by the state or jurisdiction where they occur, and the certificate the registrar issues is the document that settles the question. The widespread belief is that only immediate family can get one. In several states that is not accurate, and it is worth knowing which model a state uses before assuming you are shut out.
California issues two grades of the same certificate. Under Health and Safety Code section 103526, a requester who is not within the statute’s list of authorized persons still receives a certified copy — an informational one, bearing the printed legend “INFORMATIONAL, NOT A VALID DOCUMENT TO ESTABLISH IDENTITY.” It carries the same recorded facts. It simply cannot be used as identification.
Ohio takes a different route to a similar place. Ohio Revised Code section 3705.23 directs the state or local registrar, on a signed application and the statutory fee, to issue a certified copy of a vital record “to any applicant,” subject to the exceptions the section names. One of those exceptions is directly relevant here: for the first five years after a death, the decedent’s Social Security number is kept off the certified copy unless it is specifically requested by one of a list of people the statute enumerates — a spouse, a lineal descendant, the executor or administrator or their attorney, a county veterans service officer, and several others. Ohio decided, in other words, that the death record should be open and the number on it should not be.
Two states are two states. The models vary considerably across the country, and some jurisdictions restrict death certificates tightly to a defined class of applicants. Check the issuing jurisdiction’s own rules rather than a general article — including this one — before concluding you cannot obtain a copy. Where a request is refused, the reason given usually tells you which model that state uses and whether an informational version exists.
Six ways this search goes wrong
Each of these arrives at our desk already having cost somebody weeks or a bad decision.
Reading an empty result as proof of life
The commonest error and the most expensive. A death only enters the file if a qualifying report reached SSA, and the agency says outright that it is not keeping a comprehensive accounting of deaths. Nothing found means nothing found.
Chasing a recent death in the wrong product
A death from last year is not slow to appear. The file an uncertified buyer can reach is defined by excluding deaths inside the three-calendar-year window, so no amount of re-searching it will help.
Ordering a certificate from the wrong state
Somebody read a last-residence ZIP as a place of death and wrote to that registrar. On a current record that field is blank; on an older one it was an administrative address or the address of whoever collected a lump-sum payment.
Believing a platform that never applied the deletes
Erroneous deaths are withdrawn through delete records in the update stream. A licensee that loaded a file once and stopped keeps showing a death the federal file has already retracted, and the person it names is alive.
Settling for a single name-and-year match
Two people sharing a name and a birth year is ordinary, and the truncated name fields create matches that do not exist in the source. Without a second independent record agreeing on something else, you have a candidate.
Treating a vendor’s claim as certification
Access to current-death data requires certification by Commerce, an attested safeguarding regime and annual renewal. NTIS publishes the list of certified persons, so the claim is checkable and occasionally does not check out.
How we work a death confirmation
Four stages. The point of the order is that each one narrows the next, and none of them ends at the index.
Establish the purpose before the search
An estate that has to be administered, a benefit that has to be stopped, a title defect, a family that needs to know. The purpose is written down first, because it decides what we may lawfully do and what we will decline.
Use the index to generate candidates, not answers
Every plausible spelling, married and maiden forms, truncations, and a date-of-birth range rather than a single year. What comes back is a candidate list with the vintage of the source noted against it.
Test each candidate against something independent
An obituary naming the survivors you expect, a probate case in the right county, an interment record, a benefit termination. A candidate that survives two unrelated checks is worth the price of a certificate.
Go to the registrar, and report the disagreements
The certificate is what we are aiming at, and where the sources conflict we hand you the conflict rather than the tidier of the two readings. An unresolved file said plainly beats a confident answer that is wrong.
What this firm is, and what it will not do
Death records touch money, identity and safety at once, so the limits belong at the front rather than in a footer.
We are a skip-tracing and public-records research firm. . What we do is read records and file requests with registrars and agencies under our own name, saying who we are and who instructed us. We do not surveil, follow or photograph anyone. Every file opens with a purpose the law permits, stated in writing before any search begins — administering an estate, resolving a title defect, ending or claiming a benefit, satisfying a court’s diligent-search requirement, or a family that needs to know what happened to a relative. Curiosity about a stranger is not on that list, and neither is a request whose real object is a living person’s whereabouts dressed up as a question about a death. Where the stated purpose does not survive a few questions, we say so and decline the work rather than take the fee. Comparable records-based tracing work across other public files is described on our skip-tracing services page.
. . Death is the subject where that temptation is strongest, because grief makes people helpful and a caller claiming kinship is rarely challenged. It is also the subject where the harm is worst, because the person misled is usually somebody who has just lost someone. If a record requires a status we do not have, the answer is to go without it or to have the person who genuinely holds that status make the request. We also do not go after the contents of private financial accounts — balances, statements, account numbers, transaction histories. Those sit behind a financial institution’s door for good reason, and the usual method of getting past that door is precisely the impersonation we just ruled out.
Not a consumer reporting agency
People Locator Skip Tracing is not a consumer reporting agency and nothing we produce is a consumer report. Do not use a death confirmation or anything else traced through this page to make a decision about employment, tenancy, credit, insurance or a professional license. Those decisions are governed by the Fair Credit Reporting Act and belong with a screening firm that operates the consent, disclosure, adverse-action and dispute machinery that statute requires. We do not run that machinery. This matters more here than on most subjects, because death data flows straight into exactly those decisions — a mistaken death record has ended credit applications and job offers for living people — and if a request for a locate turns out to be one of those decisions in disguise, we decline it.
Where we stop, and why this subject in particular
A request that could be a route to a living person who does not want to be found gets more scrutiny at intake, not less. Three shapes of misuse cluster around this exact subject and we name them so nobody wastes their time. The first is the inverted question: someone asks us to confirm that a person is dead, and what they actually want is confirmation that the person is alive and a current address to go with it. Someone who has left a violent household is often deliberately hard to reach, and a query framed as bereavement is a comfortable way to ask about them. The second is the number: a record in this file carries a Social Security number, and a request whose real interest is that number rather than the death is a request we will not fill. The third is the claim — a death used as the basis for taking control of an account, an estate, a policy or a benefit by someone whose entitlement is asserted rather than documented. Where the circumstances suggest a protective order, a stalking pattern, a custody fight being run outside a court, or a relative who has already said plainly that they want no contact, we stop, and we do not restart the same file under a rewritten instruction. People live where they live, and some of them have good reasons for the distance.
Everything on this page is general information about federal and state records and is not legal advice. Whether a particular document proves a death for a particular purpose, what a court or a financial institution will accept, and who is entitled to what once a death is established are questions for an attorney in the relevant jurisdiction. We find records and show our sources; the legal conclusions belong to someone else.
Who needs a death confirmed, and what for
Six recurring instructing parties. The same record answers six quite different questions.
Stock transfer agents
Shareholders gone silent, where a death decides between reissue and escheatment.
Life settlement buyers
Holding policies on people they have never met and must not lose track of.
Charitable bequest offices
Legacy pledges that only mature on a death nobody thinks to report to them.
Title examiners
A chain of title that stalls on an owner whose death was never put on record.
Mortality researchers
Cohort follow-up that has to state honestly what its death ascertainment misses.
Families with no answer
A relative out of contact for years, and nobody willing to say either way.
The last group is the one this page is really written for. Everyone above them has a file, a deadline and a colleague to ask; a family that has lost touch with someone has a search box and a result they cannot interpret. If that is you, the single most useful thing to hold onto is that the search box is answering a narrower question than the one you asked. It is telling you whether a particular administrative report reached a particular federal system — not whether your relative is alive. Where the trail has gone cold for reasons that have nothing to do with a death, the techniques for restarting it are collected in the piece on restarting a family search that has stalled.
You get the document, or you get the reason there isn’t one
A death confirmation from us names the record it rests on: the jurisdiction, the file the entry came out of, the date the source was current, and where the sources disagree with each other. If the honest answer is that no record establishes the death, that is what the report says, in those words, with the searches we ran listed underneath it — and you are not charged for a confirmation we did not reach. Once the lawful purpose is on the record, our opening assessment — whether the product you searched could even have held that death, and which corroborating records are worth ordering — reaches you within 24 hours. A file left open is uncomfortable. A wrong death, acted on, is worse.
Questions about the index and the file behind it
Is the Social Security Death Index an official government database?
No. The federal record is the Social Security Administration’s Death Master File, built from the Numident and distributed through the Department of Commerce. “Social Security Death Index” is the label commercial genealogy and people-search sites put on their own loaded copy of an extract of that file. There is no government site by that name and no single authoritative SSDI, which is why two sites can show different results for the same person: they are showing different vintages of the same underlying data.
How soon after a death can the record be released to me?
Generally not for three calendar years. Section 203 of the Bipartisan Budget Act of 2013, codified at 42 U.S.C. 1306c, bars the Secretary of Commerce from disclosing Death Master File information about a person during the three-calendar-year period beginning on the date of that person’s death, unless the requester is certified under the Commerce program. NTIS accordingly sells two products, defined at 15 CFR 1110.2: a Limited Access DMF that includes those recent deaths and requires certification, and an Open Access DMF that excludes them and does not.
What information does a Death Master File record actually contain?
Less than most descriptions suggest. The statutory definition at 42 U.S.C. 1306c(d) is four elements: name, Social Security account number, date of birth and date of death. The NTIS record specification adds an update indicator, a name suffix and middle name, and a single character holding a proof code. Positions 82 to 100 are blank. There is no cause of death, no funeral information, no next of kin and, on any record produced after November 1, 2011, nothing geographic at all.
Does the record show where the person died?
It does not, and it never really did. The current layout has no location field: the state and county code of residence, the last-residence ZIP code and the lump-sum-payment ZIP code were removed on November 1, 2011 and those positions are always blank. On older records where values do appear, the last-residence ZIP reflected an address held in SSA’s administrative records, which may be years out of date or a care facility, and the lump-sum ZIP was where a death payment was sent, meaning the claimant’s address. Neither was a place of death.
What does the V or P code on a record mean?
It records how the death was substantiated. V means the report was verified with a family member or someone acting on the family’s behalf. P means a death certificate, or the required information from one, was observed. N or blank means the record predates those codes or came from a source considered fairly reliable without meeting either standard. Most consumer platforms never display the character, and NTIS warns subscribers not to act against any individual without further investigation even when a V or P is present, adding that SSA does not guarantee the file’s accuracy.
Why are people who definitely died missing from the index?
Because the file only knows about deaths that were reported to SSA under conditions that put them in a public product. SSA officials told the agency’s Inspector General that death information is collected to administer SSA’s programs, not as a comprehensive accounting of all deaths in the country. The July 2023 audit found roughly 18.9 million numberholders born in 1920 or earlier with no death information on their record, none of them in any version of the file. Add the state-reported deaths that section 205(r) keeps out, the three-year window on recent deaths, and derivative copies that were never refreshed, and gaps are the expected condition rather than a malfunction.
Can an incorrect Social Security Death Index record be corrected?
The federal file does have a retraction mechanism. The NTIS specification defines a delete indicator for any record previously on the file that was found to be erroneous, and a separate correction pattern in which a reversed death arrives with zeros in the date of death alongside a delete instruction. Subscribers are told to process those and remove the entry from their own copies. The practical difficulty is that the correction only reaches you if the platform you are reading applied it, so a stale commercial copy can keep displaying a death the federal file has already withdrawn. Anyone affected by an incorrect death record should take it up with SSA directly as well as with the site displaying it.
Is a match in the index enough to prove someone died?
No. A match tells you a death report exists for a person recorded under that name and date of birth. It does not tell you it is your person, and the fields that would help you tell two candidates apart were removed in 2011. Names truncate at twenty characters of surname and fifteen of given name, which creates collisions, and administrative dates of birth carry transposition errors. Treat a match as a lead, corroborate it against at least one independent record, and where it matters, obtain the death certificate from the state registrar, which is the document that actually proves the fact.
Bring us the row you found, and we will tell you what it is worth
Send the name, any dates you have, the last place you know of, and the reason you need the death established. We will tell you whether the search you have already run could even have covered that death, what a certificate is likely to cost and from where, and whether the honest answer is that the file cannot settle this — before you spend anything. Tell us what you are trying to establish and we will tell you what the records can carry.
Ask us to confirm a death
