The Defense Department’s adjudication facility published a denial figure, and almost nobody quotes it. In fiscal 2021 the DCSA DoD Consolidated Adjudications Facility closed 836,461 cases and recorded 1,810 as “Denied Revoked”: denials and revocations pooled, against a workload mixing clearances with suitability and credentialing actions. Our arithmetic on the two gives 0.22 percent.

The intelligence community’s last named-agency table, six years earlier, put the CIA at 8.5 percent and DIA at 0.0 percent. Same table, same government, same twelve months. Both are real; neither is the denial rate, because no single such number is published. When a site quotes a tidy one, ask which numerator, which denominator, which year.

Is there a current security clearance denial rate?

Not a clean one, and not nothing. ODNI’s government-wide series stopped after fiscal 2019 with its denial tables blanked. DCSA, which adjudicates the DoD population, published raw denial-and-revocation counts against its own caseload for fiscal 2021. That is the closest thing available, on a mixed basis.

DCSA’s FY21 Adjudications Year in Review, published January 2022, prints on one page: 836,461 total cases adjudicated, 1,810 denied or revoked, about 16,380 preliminary letters of intent to deny or revoke, and about 1,744 final denial or revocation letters. It explains neither letter figure, so do not read the gap as saves. Three cautions attach to the 0.22 percent: the numerator pools denials with revocations, the denominator mixes personnel security, suitability and credentialing work, and continuous-evaluation alerts and incident reports sit inside it. It bounds an order of magnitude, not an applicant denial rate.

ODNI’s series is the one that went dark. Through fiscal 2015 the reports named seven agencies and printed each one’s rates. Fiscal 2016 and 2017 printed the same columns for ten agencies labelled only Agency #1 through #10. Fiscal 2018 dropped the section, citing “new reporting requirements in the NDAA for FY 2018.” Fiscal 2019 kept the headings and blanked every value with (b)(3). Nor can a rate be rebuilt from what ODNI did print: a query of approvals “cannot distinguish between initial clearance and Periodic Reinvestigation (PR) approvals,” so the 964,138 approvals in fiscal 2019 are not 964,138 applicants.

Get the statute right; the version quoted everywhere is dead. The line usually cited, the “percentage of reviews during the preceding FY that resulted in a denial or revocation of a security clearance,” is ODNI’s section heading, tracking a subparagraph Public Law 116-92 struck out on 20 December 2019. 50 U.S.C. 3104 now asks at (b)(2)(G) for one combined percentage of investigations ending in denial or revocation, and at (b)(2)(C) and (D) for what the old text never required: favourable totals for initial adjudications and reinvestigations and, separately, the number “adjudicated unfavorably.” Congress has required numerators with denominators since 2019. The duty is to submit that report to four named committees, not to publish it, so the supportable finding is that no edition has been released publicly since April 2020, not that anyone broke the law.

What did the last published denial rates actually say?

Across the seven agencies ODNI last named, denial rates spanned nearly two orders of magnitude. Four of the seven are Defense components. What the table omits is not Defense but the DoD CAF mainstream: the military services, the Fourth Estate and cleared industry. The two columns rest on different denominators.

ODNI report on clearance determinations for fiscal 2015 (June 2016), Table 5, the last edition to name agencies. ODNI says denials come from adjudications of initial cases and revocations from adjudications of periodic reinvestigations; it never prints the base of either percentage.

Read ODNI’s explanation first. It attributes the differences to “different processes employed by those agencies”: FBI and State “may discontinue security processing due to automatic disqualifiers found during a suitability review,” and such cases “are not, therefore, categorized as security clearance denials.” An agency at 0.1 percent may be removing people earlier, under another label.

Then the scope problem, stated properly. Those four Defense agencies are not the Defense population. ODNI notes that some DoD intelligence agencies hold delegated authority only over their own civilian employees and “only conduct adjudications of contractors who already have a Top Secret clearance.” Everyone else in Defense goes to the DoD CAF. GAO reported in December 2025 that the national security population was about 5.4 million on 1 October 2022, roughly 5 million of them DoD personnel or contractors, and that DoD submits data on about 93 percent of it.

Which guidelines drive denials, and which drive suspensions?

Money leads the denial column; criminal conduct leads the suspension column. Financial considerations appeared in 378 of 486 contractor denials in 2017 and 6 of 114 suspensions. A suspension is a pause pending due process, not an ending: DSS published no outcome for the 114.

Both columns come from one signed letter, the Defense Security Service Director to the House Oversight Committee, 13 October 2017. One source, but not one period: the suspensions cover all of fiscal 2017, while the denials are whatever DOHA had issued during calendar 2017 by mid-October. The last ten weeks are not in the 486.

DSS Director’s letter to Rep. Elijah E. Cummings, 13 October 2017. Both columns count guideline citations, not rates; multiple guidelines applied to some cases; the lists cover different and unequal periods; neither comes with a population.

Say what these columns are: the 486 denials went to contractor personnel under the National Industrial Security Program who had received an interim decision between 2014 and 2016. Doubly filtered, one program, a partial year, nine years old. The letter gives no denominator, so no percentage of applicants can be built from it.

What can be computed is the internal share, our arithmetic. Dividing 378 financial citations by 486 denied cases gives 77.8 percent: the share of denied cases citing Guideline F, not the share of applicants denied over money. The nine counts sum to 691 citations across 486 cases, so some denials cited several guidelines. We run no such sum on the suspension column, whose twelve counts total exactly 114, leaving its basis unclear.

One passage is read backwards constantly. DSS had granted interim eligibility in 165 of the 486 cases, and 151 of those “were only granted because the clearance applicant had failed to disclose relevant information.” The next bullet says what that means: “Of those 165 cases, 151 people had pre-existing issues that they did not disclose and were not discovered during initial checks; the issues were not identified until the investigation was completed.” Non-disclosure explains why an interim went to someone who should not have had one. The pre-existing issue is what the final denial rested on, and the letter’s tally of those denials puts money at 378 and personal conduct, which carries falsification, at 102. Omission has its own statistic, a different one: DSS withdrew 143 interim clearances, “in the majority of these cases” on falsification or later derogatory information. See how much debt is too much, SF-86 Section 23, and how an interim clearance gets denied.

One trap if you check these names against the regulation: the codified list at 32 CFR Part 147 dates from 1998 and still prints titles retired in 2017, Guideline I as “Emotional, mental, and personality disorders,” K as “Security violations.” The operative SEAD 4 names are Psychological Conditions and Handling Protected Information.

Does a psychological condition end a clearance?

Almost never on its own. Across 2012 to 2018 the DoD CAF took 2,361,717 adjudicative actions and denied or revoked 12 for psychological issues standing alone. Two of the 486 Defense denials in 2017 cited the guideline.

Those come from a DCSA fact sheet last updated 15 October 2020. Of the 2,361,717 actions, 46,985 involved psychologically-related issues, 2.03 percent; 12 ended in denial or revocation for psychological issues alone, 0.00507 percent. A further 380 with a psychological concern plus other concerns were denied or revoked. DCSA’s words: none of the denials or revocations “were based solely on an individual seeking mental health care.” Not seeking care when it was needed, and not following the treatment recommended, were the disqualifying facts. See the SF-86 mental health questions.

Hearing data reads very differently, and the gap is a measurement artefact. The Department of Energy’s hearings office, about 140 contested personnel security cases a year from the national laboratories, records alcohol as its leading concern and psychological conditions in 29 of 202 concerns in fiscal 2023, our sum of its counts. But that ratio counts a concern raised in a case decided either way, while the Defense ratio counts a guideline cited in a case that ended in denial: a concern usually mitigated is large in the first and small in the second by construction.

How reliable is the data that does exist?

Less than it looks. GAO reviewed 305 ODNI clearance statistics for the third quarter of fiscal 2024 and found 191 of them, 63 percent, inaccurate or incomplete: 118 inaccurate, 73 incomplete.

That lands on the field a denial rate would be built from. ODNI requires agencies to report “completed adjudications,” cases adjudicated through approvals, denials and revocations. GAO found that data inaccurate or incomplete at six of the seven agencies. The National Geospatial-Intelligence Agency reported adjudicating 839 Top Secret clearances for contractors that quarter; GAO’s review of its case-level records found 142.

Both halves of GAO’s caveat belong here. It sampled seven agencies and eight reporting requirements and calls the selection not generalizable, so 63 percent is not the error rate of federal clearance data overall. In the same sentence GAO notes those seven report on more than 90 percent of all clearance holders.

What sits between adverse information and a denial?

A whole-person assessment, a possible conditional grant, and a due-process track most applicants never enter. Under 32 CFR 147.2 the adjudicator weighs nine factors, and adverse information does not automatically produce a denial.

The nine cover seriousness, circumstances, frequency and recency, age and maturity at the time, voluntariness, rehabilitation, motivation, potential for coercion, and likely recurrence. CRS restates it: adverse information “will not automatically result in denial of a security clearance.” That weighing is the whole-person concept. The regulation then states the asymmetry, and an honest page prints both halves: “Any doubt as to whether access to classified information is clearly consistent with national security will be resolved in favor of the national security.”

Between grant and denial sits an outcome most such pages omit. A DCSA fact sheet dated 1 February 2024 describes conditional eligibility, where information warrants a denial recommendation but “the specific risk to national security can be managed with appropriate mitigation measures.” It covers five of the thirteen guidelines: Sexual Behavior, Financial Considerations, Alcohol Consumption, Drug Involvement and Substance Misuse, Criminal Conduct. Personal conduct is not among them. The clock is hard: 20 calendar days from the signed cover-page receipt to return the response, or the case closes with no determination.

When a case reaches a hearing the paper trail becomes public; the outcome statistics do not. DOHA posts every industrial clearance decision individually and no aggregate at all, and the collection is not a denial log: of two dozen 2024 decisions we opened, several grant eligibility. No grant-versus-deny split exists. What arrives first is the Statement of Reasons and its 20-day clock; for people already in access, suspension under continuous vetting.

Frequently Asked Questions

What percentage of security clearances are denied?

No clean national rate is published, but numbers exist. DCSA recorded 1,810 denials and revocations against 836,461 cases adjudicated in fiscal 2021, 0.22 percent by our arithmetic, on a caseload that also contains suitability and credentialing work. The last named-agency intelligence-community rates, fiscal 2015, run 0.0 to 8.5 percent. A site quoting a current national figure is inventing it.

Which adjudicative guideline causes the most denials?

Financial considerations, in the only published breakdown of actual denials: cited in 378 of the 486 Defense contractor denials in 2017, though multiple guidelines applied to some cases. Alcohol leads a different dataset, DOE hearing concerns, at another department and stage.

Does one bad answer on the SF-86 mean automatic denial?

No. Under 32 CFR 147.2 the adjudicator applies the whole-person concept across nine factors, and CRS confirms adverse information will not automatically result in denial. Doubt resolves in favour of national security, and each case is judged on its own merits.

Are clearances denied more often for conduct or for hiding it?

No published dataset answers that. The 2017 DSS letter is often read as saying omission, because 151 of the 165 interim grants inside the 486 denied cases were “only granted because the clearance applicant had failed to disclose relevant information.” That explains why an interim was issued, not why the clearance was denied: the letter says those 151 had pre-existing issues found once the investigation finished. Of the same 486 denials, 378 cited money.

What these numbers are worth, and what they are not

The absence is narrower than it looks, and the presence weaker. ODNI has released no report on clearance determinations since April 2020, and the auditors who checked that reporting in December 2025 found the denial-bearing field unreliable at six of seven agencies. Yet DCSA published a denial-and-revocation count with a denominator, and since December 2019 the statute has required exactly that pair. The rate is less missing than unassembled.

The practical consequence is specific. Adverse endings are rare in absolute terms on the DoD side; money leads the denial column, conduct the suspension column, and a suspension is a pause, not a verdict. So file complete: write down what you hope nobody asks about, then put every item on the form in your own words, with dates. Where an item is genuinely ambiguous, a sealed record, a foreign relative, a therapy episode you are unsure Section 21 reaches, ask your facility security officer or your own counsel before you certify. The SF-86 carries criminal exposure either way. This is not legal advice, and no figure predicts an individual outcome: 32 CFR 147.2 requires each case to be judged on its own merits. Sources: DCSA FY21 adjudications, its mental-health fact sheet, GAO-26-107100, DOE OHA FY2023, 32 CFR Part 147, the 2017 DSS letter.