The reliable way to tell them apart is to look for four things at once: a control system on the material (SCI vs. collateral), the required eligibility (program access and indoctrination vs. a clearance plus need-to-know), the markings (SCI control markings and portion caveats vs. plain classification markings), and the storage and handling channel the material is built for. Any single indicator can mislead you. A document can carry a Top Secret marking and no SCI caveat and still be collateral; a document can be handled by someone with a TS/SCI clearance and still be collateral because it isn't in an SCI channel.
That's the boundary this article works inside. I'm not going to pretend there's a single stamp you can glance at and be done. In practice, separations get misjudged by experienced people for one reason: they check the marking and skip the channel, or they check the channel and skip the eligibility. You need both. What follows is the operational method β where to look, in what order, what failure looks like at each step, and the edge cases that trip up even cleared personnel.
One scope note before we go further: the specific wording of markings, systems, and storage rules lives in classified and controlled policy documents, and the exact rules change by agency and over time. Where I give a value that policy controls, I name the authority so you can confirm it against the current version rather than trusting this article. Nothing here is a substitute for your local security officer or the policy that governs your facility.
Start With the Control System and the Enabling Statute
The first split isn't a marking at all. It's which control regime the material sits under. Collateral classified material is national security information classified under Executive Order 13526 (the current order governing classification, issued in 2009) and marked Confidential, Secret, or Top Secret with the standard classification markings. Sensitive compartmented information is a separate category defined in statute β Section 4.4 of Executive Order 13526 describes it as matters concerning or derived from intelligence sources, methods, or analytical processes that are required to be handled within formal access control systems established by the Director of National Intelligence.
That statutory definition matters more than it looks. It means SCI isn't simply "more classified" than Top Secret β it's classified through a different structure. A Top Secret document and a Top Secret/SCI document can sit at the same classification level and still live in completely different worlds. The control system is the world. When you're trying to separate the two, the question you're actually answering is: which world was this material built to live in?
The practical consequence is that you can't infer the control system from the classification level. A Secret collateral document and a Secret//SCI document share a level and share nothing else. If you only record the level when you evaluate material, you've recorded the least useful fact on the page.
The other thing to internalize here: SCI exists in compartments. Each compartment corresponds to a sensitive intelligence source, method, or analytic process that has its own access list. Someone can hold SCI access for one compartment and be completely unauthorized for another that sits on the same page. That's why the separation question and the "who's allowed to see this" question are the same question asked twice.
What Program Access and Indoctrination Actually Require
Eligibility for collision is where people get this wrong most often, so slow down here. Collateral access requires three things: a favorably adjudicated background investigation supporting the right clearance level (Confidential, Secret, or Top Secret), a signed NDA, and need-to-know β a determination that the person needs the specific information to do their job. Nobody gets collateral access just because they hold the clearance. The clearance is the ticket; need-to-know is the door.
SCI access adds a fourth layer on top: program access. Before anyone can receive SCI, they must be indoctrinated into the specific SCI program β briefed on the program's handling rules, sign the nondisclosure agreement specific to that program, and be formally granted access by the program's controlling authority. Indoctrination isn't paperwork theater. It's the mechanism by which the program's access list exists at all. You can't be "accidentally" read into an SCI program the way you can be accidentally copied on a collateral email.
That asymmetry is the practical identification tool. If a person or a workspace handles the material without an SCI indoctrination record, the material isn't SCI β or someone is out of bounds. There's no third option. This is one of the cleanest separators available, and it's the one most often skipped because it requires asking an uncomfortable question: who, specifically, is read in?
The failure case is worth naming because it happens: material gets handled by a properly cleared person who lacks program access. That doesn't make the material collateral. It makes the handling a security incident. If you're classifying material and the only route to the room is "someone with a TS clearance carried it," check whether that person's indoctrination covers the program. If it doesn't, the material never should have been in that room, and the separation question is moot until the incident is closed.
One boundary condition: program access is granted per program, and a person can be in several. A single indoctrination record settles access to that program and nothing else. When you review eligibility for material that spans compartments, verify each compartment separately β one blanket "he's read in" covers exactly one program.
Read the Markings in the Order They Govern
Markings are where identification becomes mechanical, and the order you read them in matters. Start with the overall classification line at the top and bottom of each page. Then check for the SCI control marking β banners and portion markings that identify the SCI control system and, where applicable, the compartment. Then look at the portion markings within the document, because a document can have SCI portions and collateral portions, and the overall banner reflects the highest. The document's control system is determined by whether SCI control markings appear, not by the banner level.
The pattern to expect on a pure SCI document: a classification level, a double slash, and the SCI control identifier, with compartment indicators following where the program requires them. A collateral document carries the classification level and, as applicable, dissemination control markings like NOFORN or REL TO β but no SCI control marking. The double-slash structure is the visual tell, but only because it encodes the control system. Don't treat the slashes as the fact. Treat them as the pointer to the fact, then verify the fact.
Here's the comparison that most guides never build, because it forces you to hold both columns at once:
| Dimension | Collateral classified material | SCI classified material |
|---|---|---|
| Governing authority | Executive Order 13526 | EO 13526 Β§4.4 plus DNI-directed control systems |
| Classification levels used | C, S, TS | C, S, TS within the SCI control system |
| Required eligibility | Clearance + signed NDA + need-to-know | Clearance + NDA + need-to-know + program access/indoctrination |
| Marking signature | Classification level plus dissemination controls | Classification level plus SCI control marking and portion caveats |
| Storage channel | Approved container for the level | SCI-approved storage and handling per program rules |
| Who can be told it exists | Anyone with need-to-know at the level | Only those indoctrinated into the program |
| Common misjudgment | Assuming TS means SCI-eligible | Assuming the level alone identifies the system |
Read the table left to right and the separation becomes structural rather than cosmetic. The marking column is the fastest screen; the eligibility and storage columns are the confirmation. If the marking says SCI but no one on the handling chain has program access, you have a marking problem or an access problem β not collateral material.
The trap worth flagging: dissemination control markings. NOFORN, REL TO, and similar controls restrict who may receive collateral material, and they look restrictive enough that people read them as compartmentation. They aren't. Dissemination controls limit distribution within the collateral world; they do not create program access or an SCI channel. A NOFORN document can be entirely collateral. If you're separating on "how restricted does this look," you'll classify a lot of collateral as SCI and miss the actual SCI sitting next to it.
Use the Storage Channel as Your Confirming Test
The channel test is the one to reach for when the markings are ambiguous, the document is a working draft, or you're evaluating a space rather than a page. Collateral classified material is stored in containers approved for its level under the applicable storage rules, and access is controlled by need-to-know. SCI material is stored and handled through SCI channels β approved spaces, containers, and handling procedures established for the specific program. The physical and procedural infrastructure is different because the access model is different.
Practically, this means you can often identify the system by asking what the container was built for. A GSA-approved container rated for Top Secret collateral is rated for collateral Top Secret; it does not, by itself, authorize SCI storage. If SCI material turns up in a container that was only ever approved for collateral, the material hasn't changed systems β someone has violated the storage rule. That distinction is the whole reason the channel works as a test: the container's approval follows the control system, and mismatches are evidence of an error, not evidence of reclassification.
Storage rules also give you the cleanest edge case in the whole topic: the classified working draft. A draft that quotes SCI source material and hasn't yet been marked carries SCI content in an unmarked form. Ask the channel question β where was it produced, who handled it, what system will it be entered into β and the answer usually settles it. Drafts get misjudged because people mark from the page and the page is incomplete. Mark from the channel and the draft comes out right.
One more channel-level failure worth naming: derivative classification. When you extract from an SCI source into a new product, the derivative carries the SCI markings of its source, not the markings of the product's intended audience. If someone "downgrades" an extract by dropping the SCI caveat to share it more widely, that's an unauthorized removal of a control marking, and the material remains SCI. The channel is where you catch this, because the source document is still sitting in the SCI system and the extract in someone's collateral folder is the anomaly.
Apply the Two-Reviewer Test Before You Sign Anything
Single-reviewer identification is where most real-world errors enter, so build the second check into the process rather than treating it as bureaucracy. The two-reviewer test is simple: reviewer one determines the control system from markings and channel; reviewer two independently verifies eligibility and storage. Neither reviewer sees the other's conclusion before recording their own. Divergence between them is the signal β it means the material sits on the boundary and needs a security officer, not a coin flip.
The reason independence matters is anchoring. If reviewer two reads reviewer one's "SCI" conclusion first, reviewer two will find reasons to agree, and the second review becomes a signature rather than a check. Independent recording costs you five extra minutes per item and catches precisely the cases the single reviewer misses: the document with SCI portion markings but a collateral banner, the container approved for the wrong level, the handler without indoctrination.
Run it as a sequence, not a conversation:
1: Reviewer one reads the overall banner, the portion markings, and the dissemination controls; records the control system indicated by markings only. 2: Reviewer one checks the storage location and container approval against the level indicated. 3: Reviewer two, independently, pulls the eligibility record of every person on the handling chain and confirms program access for each compartment indicated. 4: Reviewer two records the control system supported by eligibility and channel, without reference to step one. 5: Compare the two records. Agreement closes the item; disagreement goes to the security officer with both records attached.
The failure handling is the part people skip. When the reviews disagree, don't resolve it by seniority or by whichever reviewer seems more confident. Quarantine the material in the more restrictive channel while the dispute is open, because the cost of over-restricting for a day is trivial next to the cost of an SCI spill. Then have the security officer determine the system and document the determination, so the next reviewer inherits a decision rather than repeating the argument.
The two-reviewer test also scales to workspaces, which is where it earns its keep. Review a room by sampling its containers and its access list in the same paired way: one reviewer on markings and storage, one on who's read in. A room where every container is SCI-approved but half the occupants lack program access is an SCI space with an access problem, and you've found it without opening a single document.
Know Where SCI Lives Even When the Page Doesn't Say So
Some SCI material is identifiable by origin rather than by marking, and this is the angle most identification guides skip entirely. SCI content is generated inside SCI channels β sensitive intelligence reporting, analytic products derived from compartmented sources, and the working files of indoctrinated personnel. If you know the producing office and the channel, you can identify the system from provenance when the markings are absent or degraded.
Provenance matters most with legacy material. Documents produced under earlier classification guidance may carry older marking conventions, and the current marking scheme doesn't map cleanly onto them. A legacy document with a nonstandard SCI indicator is still SCI if its source channel was SCI. When in doubt, the system of record wins over the page. That's a rule worth stating explicitly because it inverts the instinct to trust the document in front of you: the page is a report about the material, and reports can be wrong.
The corollary concerns what people can be told. SCI material's very existence is program information. Someone without program access doesn't just lack permission to read the document β they lack permission to be told the program exists or that the document exists. That's a qualitatively different boundary from collateral need-to-know, and it's why mishandling SCI is treated as more serious than mishandling collateral at the same level. A spill of collateral Secret is a containment problem. A spill of SCI can expose a source or method, which is a different order of harm.
Compartmentation accumulates, which is where identification gets genuinely hard. A document can carry SCI markings for one compartment and derivative information from another, and the access required is the union of the compartments indicated β not the highest-looking one. Reviewers who check "is this person SCI-cleared" and stop will clear someone for a document that requires a compartment they don't hold. Check each compartment on the page against the individual's record. The union is the requirement.
Do the Eligibility Arithmetic on the Handling Chain
Here's an original computation you won't find in a typical marking guide, built entirely from the access requirements stated above. Take any document and trace the handling chain: every person who has touched it, from originator to current holder. Count them. Now, for a collateral document at Secret, each person on that chain needs exactly two gates satisfied: a Secret or higher clearance, and need-to-know. For an SCI document, each person needs three: clearance, need-to-know, and program access for every compartment marked on the page.
The arithmetic does the arguing for you. A twenty-person handling chain for collateral material requires twenty satisfied two-gate checks. The same chain for SCI material requires twenty satisfied three-gate checks, plus a program access record for each compartment on each person. If you find any node in the chain where the third gate is unmet, the material was mishandled β and if you find that the third gate is unmet across the board, the likelier explanation is that the material isn't SCI and someone over-marked it. Either way, the arithmetic resolves the ambiguity that markings alone leave open.
This is also how you audit a program's boundary without reading its contents. Pull a sample of documents from a container, list the persons on each chain, and check gate three against the program's access roster. The tally gives you the separation status of the whole container in an afternoon, and it catches both directions of error: SCI in collateral channels and collateral sitting behind SCI restrictions it never needed.
The cost side is worth a sentence because it's what makes organizations skip the check. Verifying program access for a handling chain is slow β it's record lookups, not a glance. Facilities that skip it trade an afternoon of verification for a spill investigation that costs weeks and can end careers. The arithmetic is the argument for spending the afternoon.
Distinguish Compartmentation From Collateral Dissemination Controls
The most persistent confusion in this whole area is between compartmentation and dissemination control, and it deserves its own section because getting it wrong invalidates every downstream judgment. Dissemination controls β NOFORN, REL TO specific countries, and related restrictions β limit who may receive collateral classified information. They are distribution constraints on material that remains collateral. Compartmentation is different in kind: it's the formal access control system that makes information SCI in the first place, with its own access list, its own indoctrination, and its own storage channel.
The practical test: if removing the restricting marking would make the material shareable with any cleared person with need-to-know, it was a dissemination control. If removing the marking would require a program access grant before sharing, it was compartmentation. The first is a fence around the yard; the second is a different yard. Reviewers who conflate them over-restrict collateral and under-protect SCI, which is the worst of both errors.
Neither category substitutes for classification level, which is a third and separate axis. A document can be Secret, NOFORN, and collateral β three facts, one document, all consistent. Add an SCI caveat and you've changed the control system without necessarily changing the level. That's why identification has to record all three axes: level, dissemination controls, and control system. Record two and drop one, and the record is wrong in a way that won't show up until the material moves.
The failure case is a reviewer who "downgrades" a document by removing a dissemination control to widen distribution, assuming the controls were the sensitive part. They weren't. If the document was SCI, the caveat wasn't optional, and removing it created an unauthorized disclosure path regardless of who received the document afterward. Dissemination controls are negotiable by policy authority; compartmentation is not negotiable by anyone below the program's controlling authority.
Recognize the Mistakes That Produce Misidentification
Most misidentification isn't ignorance of the rules; it's a small set of recurring errors applied confidently. The most common is marking-blindness: reading the classification level, seeing "Top Secret," and assuming that settles the system. It settles the level and nothing else. The second most common is the credential fallacy: inferring the material's system from the holder's clearance. A TS/SCI clearance holder can hold collateral material and a TS collateral holder cannot hold SCI β the clearance tells you about the person, not the page.
A third error is container inference in reverse: assuming that because material is stored in a restricted room it must be SCI. Restricted rooms control access to collateral material too. The container's approval rating is the fact; the room's general restrictiveness is not. And a fourth is banner dependence for compound documents β trusting the overall banner on a mixed document instead of reading the portion markings. A document whose banner says Top Secret and whose portions include both SCI and collateral content is an SCI document with collateral portions, and the portions carry the markings that govern their own handling.
The single most expensive error, though, is quiet and procedural: dropping the caveat to make sharing easier. It's easy to do, easy to rationalize, and it converts a handling question into a spill the moment it happens. Every skill in this article is downstream of a simple discipline β when the markings and the channel disagree, stop and treat the material as the more restrictive of the two until a security officer resolves it. Over-restriction is recoverable in an afternoon. The alternative isn't.
FAQ
Is all Top Secret information SCI?
No β most Top Secret information is collateral, and the classification level alone never establishes SCI. SCI is defined by the control system and program access requirement, not by the "Top Secret" label. A Top Secret document without an SCI control marking and without an SCI handling channel is collateral at the Top Secret level. Assuming the reverse is the most common misidentification in the field.
Can collateral and SCI portions appear in the same document?
Yes, and the governing rule is portion-by-portion. A document can carry SCI markings on some portions and collateral markings on others, with the overall banner reflecting the highest level. Each portion is handled according to its own marking, and the document as a whole lives in the more restrictive channel. Reviewers who read only the banner will misroute the collateral portions and, worse, lose track of the SCI ones.
What happens if SCI material is stored in a container approved only for collateral?
That's a storage violation, not a conversion β the material remains SCI and the container's approval hasn't changed. The correct response is to stop handling, secure the area, and notify the security officer, who will direct the containment. The container being "approved for Top Secret" is not a defense, because storage approval follows the control system, not the classification level.
Does a TS/SCI clearance mean someone can see any SCI document?
No. An SCI clearance establishes eligibility to be considered for program access; the access itself is granted per program through indoctrination. A person cleared at TS/SCI but not indoctrinated into the compartment on the page cannot see that material. When a document carries multiple compartments, the requirement is the union of all marked compartments, not the highest or the most familiar one.
How long does it take to verify a handling chain's program access?
Expect an afternoon of record lookups for a container-sized sample, because each person on each chain must be checked against the program's access roster. That's slow by design β the records are the access control system. Organizations that skip the check trade the afternoon for a spill investigation measured in weeks, which is why the paired two-reviewer method pays for itself the first time it catches a single undocumented handler.
Are NOFORN and REL TO markings a form of compartmentation?
No. Dissemination controls restrict who may receive collateral classified information; they don't create program access or an SCI channel. The distinguishing test is what happens if the control is removed by the proper authority β collateral becomes shareable with cleared need-to-know holders, while SCI still requires program access. Treating NOFORN as compartmentation over-restricts collateral and obscures real SCI sitting beside it.
Is SCI always classified at a higher level than collateral?
Not necessarily. SCI appears at Confidential, Secret, and Top Secret levels within its control system, and a Secret//SCI document carries the same level as a Secret collateral document. The difference is the control system, not the level. That's precisely why level alone can't answer the separation question, and why every identification record needs to capture control system and dissemination controls alongside classification.
This article is general educational information about security classification concepts, not a legal opinion or an authoritative statement of handling rules. The specific marking conventions, storage standards, and program access procedures are controlled by classified and agency policy and change over time; confirm them with your facility security officer or the governing authority before acting. If you encounter material you believe is mishandled, stop handling it and report through your organization's security channels rather than resolving it yourself.