Ethernet cable color codes: the wires inside, the jackets outside
Open a network bundle and the question writes itself. There is a red line in here, a fistful of purple ones, some green, some white, blue, yellow. Why so many colors? When do you use which? What does each one mean? The honest answer is that two completely different color systems are being looked at as if they were one. The eight tiny wires inside a single cable follow a real wiring standard — their colors are load-bearing and you cannot change them. The color of the jacket outside — the blue, red, yellow, purple you see in the rack — is a convention somebody chose, not a standard, and a blue cable is no faster, no better, and no higher-category than a grey one. This article separates the two, because almost every wrong belief about cable color comes from mixing them up.
Two color systems wearing the same word.
When someone points at a bundle and asks why there are so many colors, they are usually looking at two answers at once and can't tell them apart.
The first system is inside every single cable. Strip the jacket off any Ethernet cable and you find eight thin wires twisted into four pairs, colored blue, orange, green and brown, each pair made of one solid-color wire and one white wire with a matching stripe. That coloring is not decorative and it is not optional — it is a published wiring standard (ANSI/TIA-568) that tells the installer exactly which wire goes to which of the eight pins in the RJ45 plug.¹ Get it wrong and the cable does not work, or works badly. This is the system people mean when they say “RJ45 color code.”
The second system is outside, and it is the one that produces the visible riot of color in a rack: a red patch cord here, a bundle of purple ones there, blue for most things, a couple of yellow. That is the jacket color, and — this is the part that surprises people — there is no universal standard behind it for copper cabling. The electrical performance of a cable is identical whatever color the jacket is.¹⁰ The red one is red because somebody on that site decided red would mean something. On a different site it means something else, or nothing.
Nearly every myth about cable color — “blue is Cat 5, the colored ones are faster,” “red means it's the important one,” “yellow is always PoE” — comes from treating the convention on the outside as if it were the standard on the inside. The rest of this article takes the two apart: first the wires (where the rules are real), then the jackets (where the rules are local), then how a working site actually decides what to use.
Four twisted pairs, eight wires, and why one of each pair is striped.
A standard Ethernet cable carries four twisted pairs — eight conductors. The pairs have base colors of blue, orange, green, and brown. Each pair is two wires: one a solid color, the other white with a stripe of the same color — white/blue with blue, white/orange with orange, and so on.¹ The two wires of a pair are twisted around each other along the whole length of the cable, and the twisting is the entire point: the two wires carry the same signal in opposite polarity, so any interference that hits both of them cancels out. The tighter and more consistent the twist, the better the cancellation and the higher the data rate the cable can carry.³
The solid-plus-striped convention is not arbitrary either — it is inherited directly from a century of telephone wiring, where the two wires of a pair were called “tip” and “ring” after the parts of the operator's switchboard plug, and were distinguished by a major color and a minor (stripe) color.⁷ Ethernet kept the scheme because it works: it lets an installer tell the two halves of a pair apart at a glance while keeping them visibly grouped.
Those eight wires land on the eight pins of an RJ45 (8P8C) connector in a fixed order defined by one of two patterns, T568A or T568B. Here is the full mapping. The pair column and the role column are shown for T568B (the more common pattern in North America — more on that next):
| Pin | T568A | T568B | Twisted pair | 10/100 role |
|---|---|---|---|---|
| 1 | White/Green | White/Orange | Orange pair | Transmit + |
| 2 | Green | Orange | Orange pair | Transmit − |
| 3 | White/Orange | White/Green | Green pair | Receive + |
| 4 | Blue | Blue | Blue pair | Unused |
| 5 | White/Blue | White/Blue | Blue pair | Unused |
| 6 | Orange | Green | Green pair | Receive − |
| 7 | White/Brown | White/Brown | Brown pair | Unused |
| 8 | Brown | Brown | Brown pair | Unused |
The single most important thing this table shows is hiding in the pair column: pins 3 and 6 are one pair, even though pins 4 and 5 sit physically between them. The green pair (in T568B) is split across non-adjacent pins by design, so that the connector's pin geometry keeps each pair's two wires electrically balanced.¹ An installer who lays the eight wires into the connector in a naive flat rainbow — pin 3 to one pair, pin 4 to the next, ignoring which wires are actually twisted together — will pair pin 3 with pin 4 and break the cable in a way that section 04 explains.
Why there are two patterns, which to use, and why it barely matters which.
Compare the T568A and T568B columns above and the difference is exactly one swap: the orange pair and the green pair trade places on pins 1-2 and 3-6.¹ The blue pair (4-5) and brown pair (7-8) are identical in both. That is the whole difference. Electrically, A and B are equal — a cable wired A-to-A performs exactly the same as one wired B-to-B.²
Two patterns exist for historical reasons, not technical ones. T568A follows the older USOC telephone wiring scheme and stays backward-compatible with one- and two-pair phone wiring; T568B preserves the pattern used by AT&T's 258A (Systimax) system, which was the dominant commercial wiring in the United States when the standard was written, so it was folded in as a second option.¹ Because AT&T was so widespread, T568B became the de-facto default for North American commercial and residential work, and it is what most equipment and most technicians expect to see.² T568A historically carried a recommendation for US federal contracts and residential installs, though current revisions of the standard have softened that.²
So which should you use? Either — but pick one and use it everywhere on the site. For a typical US UniFi or structured-cabling job, T568B is the path of least surprise. The only rule that genuinely matters is consistency: both ends of any one cable must use the same pattern. Terminate one end A and the other end B and you have unintentionally built a crossover cable — which leads directly to the next point.²
The crossover cable, and why you can forget it exists
A crossover cable is deliberately wired T568A on one end and T568B on the other, swapping the transmit and receive pairs.⁵ For years it was how you connected two of the same kind of device directly — PC to PC, or switch uplink to switch uplink — because both ends would otherwise try to transmit on the same pins. Modern gear made it obsolete. Every gigabit (and faster) port implements Auto-MDI-X, which detects the situation and swaps transmit and receive internally, so the cable type no longer matters; gigabit links also use all four pairs in both directions, which removes the concept entirely.⁵ The practical upshot: you almost never need a crossover cable today, and an accidental A/B mismatch on a modern link often still connects — which is exactly why it is a documentation hazard rather than an obvious failure. Build your cables straight-through, consistently, and the problem disappears.
The split pair: how a cable passes the cheap test and still fails.
Here is the mistake that the inside-the-cable color code exists to prevent. It is possible to land all eight wires on the correct pins at both ends — pin 1 to pin 1, pin 2 to pin 2, every conductor continuous end-to-end — while still pulling the wrong two physical wires into a signalling pair. This is a split pair: the pin map is right, but pins that must carry a balanced differential signal together (3 and 6) are taken from two different twisted pairs.⁴
The reason this is so dangerous is what it does to testing. A $20 cable mapper or a continuity beeper only checks that each pin connects to the matching pin at the far end. A split pair passes that test cleanly — all eight conductors light up green — because every pin is connected to the right pin. What the cheap tester cannot see is that the two wires now sharing a signal are no longer twisted around each other, so they have lost their noise cancellation. The result is severe near-end crosstalk (NEXT), which a certification tester flags but a continuity light never will.⁴ In Fluke's words, a split pair “will pass a standard continuity test, but will have serious crosstalk problems, and will most likely not perform adequately at specified data rates.”⁴
Whether the fault bites also depends on speed, which is why it can hide for years. 10BASE-T and 100BASE-TX use only two pairs — transmit on pins 1-2, receive on pins 3-6 — and leave the blue and brown pairs idle.³ Gigabit (1000BASE-T) and everything above it use all four pairs simultaneously in both directions.³ So a sloppily terminated drop can run 100 Mbps for years, then the moment a gigabit device is plugged in — a new access point, a NAS, a 2.5 GbE camera — the link renegotiates down, throws errors, or flaps, and it looks like the new device is faulty when the cable was wrong all along.
The lesson is the one the color code is built to teach: follow the pairs, not a flat 1-to-8 rainbow. Keep each twisted pair twisted right up to the connector, untwisting only the last few millimetres, and lay the wires so that pins 3 and 6 come from the same pair. “It lit up green on the tester” is not proof a cable is right — it is proof the eight conductors are continuous, which is a much weaker claim.⁴ (The mechanical side of this — jacket-into-the-strain-relief, the 13 mm untwist limit, certification — is covered in our cabling and patch-hygiene article.)
What the conductors carry when PoE is on.
Power over Ethernet rides on the same eight conductors, which is one more reason pair integrity matters. There are two ways power is injected. Mode A puts the DC on the data pairs — pins 1-2 and 3-6 — superimposed on the signal. Mode B puts it on the two pairs that 10/100 left idle — pins 4-5 and 7-8.⁶ The original PoE standards (802.3af, and 802.3at / PoE+) deliver power over two pairs; the newer 802.3bt (PoE++, Type 3 and 4) uses all four pairs to reach up to roughly 51 W and 71 W at the device.⁶
Under four-pair PoE++, every conductor in the cable is now carrying both data and DC current. A marginal termination or a split pair is no longer just a data-balance problem — it becomes a current-handling and heat problem too. The same care that keeps a gigabit link clean is what keeps a powered camera, doorbell, or high-power access point stable. None of this changes the wire color code; it just raises the cost of getting it wrong. (For sizing the power budget itself, see our PoE budgeting article.)
The jacket colors are a convention, not a standard.
Now the outside — the colors you actually see in a bundle. The first thing to understand is the thing most people get backwards: for copper Ethernet, there is no universal standard that assigns meaning to jacket color. No ANSI/TIA rule says blue is data and red is security. A cable performs identically regardless of its jacket color, so the color is chosen for human convenience, not for the electronics.¹⁰
That is genuinely useful — colour lets a technician see at a glance what a cable is for, avoid yanking the wrong one, and trace a service across a crowded rack — but the meaning is entirely local. The conventions that float around the industry exist, and they overlap, but they also openly contradict each other. Depending on which vendor guide you read:
- Blue is the one near-universal agreement: standard data / user / horizontal connections.¹⁰¹¹
- Red means VoIP and emergency lines in one common scheme — and critical links, firewalls, or security in another. Directly conflicting.¹⁰¹¹
- Yellow is PoE in most schemes, but also shows up for VoIP, alarms, or cameras.¹⁰¹¹
- Green means crossover in the legacy scheme, but security cameras / access control in modern enterprise schemes.¹⁰¹¹
- Orange tends to mean the demarcation / carrier hand-off, echoing the one formal standard below; elsewhere it is tagged to analog or voice.⁹¹¹
- Purple, grey, white, black, pink are general-purpose or reserved for whatever a given site needs (management, guest, DMZ, A/V, outdoor).¹⁰¹¹
Because the schemes disagree, you cannot walk into an unfamiliar building and read its colors with confidence. A red cable means whatever that site documented it to mean — and if nobody documented it, it means nothing at all.
TIA-606: a color code for the labels, not the jackets.
There is one place a formal color code does live on the outside of the plant — but it is an administration standard, not a wiring one. ANSI/TIA-606, the Administration Standard for Telecommunications Infrastructure, ties together identifiers, labels, records and a recommended color code that identifies the type of termination at a cross-connect or patch field.⁸ Critically, these colors were conceived for the termination fields, backboards and labels — the patch-panel sections and the tags — not for the cable jacket itself, and the standard explicitly treats color as recommended, not mandatory.⁸ The recommended assignments:
| Color | TIA-606 termination type |
|---|---|
| Orange | Demarcation point / central-office termination |
| Green | Network connections, customer side of the demarc |
| Purple | Common equipment (PBX, LAN, computers) |
| White | First-level backbone terminations |
| Gray | Second-level backbone terminations |
| Blue | Horizontal cabling (closet to work-area outlet) |
| Brown | Inter-building (campus) backbone |
| Yellow | Auxiliary, alarms, security, maintenance |
| Red | Key telephone systems |
This table is the ancestor the informal patch-cord conventions borrow from — it is no accident that “blue = data” is the one everyone agrees on, since TIA-606 blue is the horizontal cabling subsystem, which is the LAN drop to the desk.⁹ But the mapping is loose: the by-service jacket schemes (data / voice / PoE / cameras / guest) layer organizational meaning on top of, or instead of, the TIA-606 subsystem meaning. The two are not the same thing, and conflating them is how the contradictions in section 06 arise.
RED/BLACK: the one context with a hard security meaning.
There is exactly one widely-cited environment where cable color — or more precisely, the RED and BLACK designation — carries a binding meaning: US-government and defense facilities operating under TEMPEST doctrine. In the RED/BLACK concept, RED identifies systems, equipment and cabling that carry unencrypted, classified (plaintext) information, and BLACK identifies those that carry encrypted (ciphertext) or unclassified information.¹²
The rule is not aesthetic. RED and BLACK paths must be kept physically separated and/or shielded so that compromising electromagnetic emanations from a plaintext (RED) circuit cannot couple into a BLACK one and leak classified data. The separation, identification and installation requirements are set out in US-government guidance — CNSSAM TEMPEST/1-13, “RED/BLACK Installation Guidance” — and administered through a Certified TEMPEST Technical Authority.¹² It is the exception that proves the rule for everyone else: when color/identity genuinely must mean something, an actual standards body writes down exactly what it means and enforces it. Outside of that world, your colors mean whatever your own documentation says.
On fibre, the jacket color actually tells you something.
One important asymmetry, because it is easy to assume copper and fibre work the same way: they do not. On fibre, jacket color is close to a real indicator of type, governed by TIA-598. The common non-military assignments: yellow for single-mode (OS1/OS2), orange for OM1/OM2 multimode, aqua for OM3/OM4 laser-optimized multimode, and lime green for OM5.¹³
So an aqua patch cord really does suggest OM3/OM4 multimode in a way that a blue Ethernet cord does not reliably suggest anything. Even here it is a strong convention rather than a guarantee — TIA-598 allows other colors if the jacket print identifies the fibre type, and the table is for non-military use — so the printed legend on the jacket still wins over the color. But the contrast is worth keeping straight: on copper, jacket color is decoration plus local convention; on fibre, jacket color is a meaningful (if not absolute) type indicator.¹³
Color has nothing to do with category.
The most expensive misconception in this whole topic: a cable's jacket color tells you nothing about its category, shielding, or gauge. A blue cable is not Cat 6 because it is blue. A blue Cat 5e cable and a blue Cat 6A cable look identical from the outside.¹⁴ Color is arbitrary; category is a property of how the cable is built and is proven two ways only — the printed text on the jacket (manufacturers print the category designation, construction and ratings along the sheath), and a field certification test that measures the cable against the category limits.¹⁴
This matters on the buying side and the auditing side. A bulk reel sold as “Cat 6” in an attractive color is not Cat 6 because of the color or even because of the box; it is Cat 6 if the construction and a certification test say so. Read the jacket print, not the jacket color. And do not confuse the inside pair colors (the T568A/B scheme, which is fixed) with the outside jacket color (which is generic) — they are unrelated systems, as this whole article has been at pains to separate.¹⁴
One adjacent non-color decision worth flagging here: shielded versus unshielded. That choice is about grounding and the electromagnetic environment, not about color, and a shielded cable whose shield is not properly bonded to ground at the terminations can perform worse than plain unshielded cable. Color conventions should never be read as implying shielding.
Pick one wiring standard, document the jacket scheme, label both ends.
Pulling the two systems back together into a working practice:
- Inside: choose T568B and use it everywhere. For a typical US install, T568B is the expected default; T568A is fine too if a spec calls for it. The non-negotiable is consistency — same pattern on both ends of every cable, every jack, every patch panel, across the whole site.² Keep the pairs twisted to the connector; never split a pair.
- Outside: a jacket color scheme is only as good as its documentation. Color is a fast visual cue; the source of truth is the label and the record. Per TIA-606, label both ends of every cable with mechanically-printed (not handwritten) labels, keep a legend in the rack/IDF, and keep an as-built record that maps every run.⁸ A color scheme nobody wrote down degrades to noise the moment the person who invented it leaves.¹¹
- A sensible starter scheme, to be adopted and then written into the legend — not treated as universal: blue for standard data, a distinct color for voice, yellow for PoE devices (APs and cameras), a reserved color for uplinks/trunks, another for management/out-of-band, red for critical or security, and a separate color for an isolated guest VLAN. The exact mapping is the site's choice.¹¹
- Scale the effort to the site. On a small home install, full jacket color coding is mostly overkill — the defensible exceptions are making PoE runs and the WAN/uplink obvious. In multi-tenant buildings, data centres and security-sensitive sites, a documented scheme is near-essential and pays back in faster, safer moves-adds-changes.¹¹
- Drop dead conventions. Reserving a color for “crossover” is wasted today — Auto-MDI-X made crossover cables obsolete, so free that color for a live service.⁵
The thread through all of it: the colors inside the cable are a standard you must obey, and the colors outside the cable are a language your site invents and then has to write down. Treat the first as fixed and the second as documentation, and the bundle stops being a mystery.
Where this article simplifies.
- A and B being “equal” assumes consistency. T568A and T568B are electrically identical only when each cable uses one pattern on both ends. The whole risk is the mixed cable, which modern Auto-MDI-X often papers over — making the defect a documentation and troubleshooting problem rather than a dead link.²⁵
- The TIA-606 table is recommended, not enforced. Secondary sources phrase a few rows slightly differently (purple as “common equipment” vs “PBX,” white/grey backbone wording), and the standard treats color as voluntary guidance layered on top of mandatory labelling and records.⁸⁹
- The jacket conventions are real but not canonical. The by-service colors in section 06 are drawn from vendor and data-centre guides that genuinely disagree with each other. They are presented as the landscape, not as a scheme to copy verbatim.¹⁰¹¹
- RED/BLACK is a separation doctrine, not a “paint it red” rule. In TEMPEST practice RED and BLACK designate security domains enforced by physical separation, shielding and labelling; the point here is that this is the one place identity carries a binding, standards-body-defined meaning — not that a jacket must literally be a given color.¹²
- Even fibre color is an indicator, not a certificate. TIA-598 colors strongly suggest fibre type, but the standard permits exceptions when the jacket print identifies the type, and applies to non-military use. The printed legend is still the authority.¹³
None of this changes the two load-bearing facts: inside the cable the color code is a standard you follow to the pin, and outside the cable the color is a convention your site defines and documents. For a network whose bundle is a tangle of undocumented colors that nobody can read, a labelling and documentation pass turns it back into infrastructure — a legend in the rack, printed labels on both ends, and an as-built record handed back with the work.
References [14]
- [1]Wikipedia — ANSI/TIA-568. Secondary summary of the U.S. structured-cabling standard: the full T568A and T568B pin-to-color mappings, the four twisted pairs (blue, orange, green, brown) and their pin assignments, the fact that pins 3 and 6 are one pair, and the statement that “pairs 2 and 3 (orange and green) are exchanged” between A and B. en.wikipedia.org — ANSI/TIA-568
- [2]trueCABLE Cable Academy — T568A vs T568B. Installer-vendor reference: pin-by-pin color order for both patterns, the orange/green swap at pins 1-2 and 3-6, T568B as the North-American default, the T568A federal/residential history, the rule that the two are identical if used consistently, and the warning that mixing A and B ends creates a crossover. truecable.com — T568A vs T568B
- [3]Wikipedia — Ethernet over twisted pair. Source for pair usage by speed: 10BASE-T and 100BASE-TX transmit on pins 1-2 and receive on 3-6 (two pairs, with 4-5 and 7-8 idle), while 1000BASE-T and faster use all four pairs bidirectionally; and for the role of the twist in noise cancellation. en.wikipedia.org — Ethernet over twisted pair
- [4]Fluke Networks Knowledge Base — Split Pair Identification. Authoritative on the split-pair fault: a split pair “will pass a standard continuity test, but will have serious crosstalk problems, and will most likely not perform adequately at specified data rates,” and is found by a wiremap/certification test rather than a bare continuity check. flukenetworks.com — Split Pair Identification
- [5]Wikipedia — Ethernet crossover cable. The crossover cable defined as T568A on one end and T568B on the other, its historical use to connect like devices, and its obsolescence: Auto-MDI-X is implemented on essentially all gigabit interfaces, and gigabit links use all four pairs, so crossover cables are no longer required. en.wikipedia.org — Ethernet crossover cable
- [6]Wikipedia — Power over Ethernet. Source for PoE Mode A (power on the data pairs, pins 1-2 and 3-6) versus Mode B (power on the spare pairs, pins 4-5 and 7-8), and for 802.3af/at being two-pair while 802.3bt (Type 3/4, PoE++) optionally uses all four pairs for up to ~51 W and ~71 W at the device. en.wikipedia.org — Power over Ethernet
- [7]Wikipedia — 25-pair color code and Tip and ring. Origin of the solid-color + white-stripe (major/minor color) convention in telephone wiring, and of “tip” and “ring” as the two conductors of a pair, named for the manual switchboard plug — the heritage Ethernet inherited for its pair coloring. en.wikipedia.org — 25-pair color code
- [8]Cabling Installation & Maintenance — ANSI/TIA-606-B coverage. The administration standard that ties identifiers, labels, records and a recommended color code together; color is “recommended but not a requirement,” applied to termination fields and labels, with each printed identifier linked to a record. cablinginstall.com — ANSI/TIA-606-B
- [9]NavePoint — Cable Color Codes: ANSI/TIA/EIA-606. Plain-language table of the TIA-606 recommended color-to-termination-type assignments (orange = demarc, green = customer-side network, purple = common equipment, white/grey = backbone tiers, blue = horizontal, brown = inter-building, yellow = auxiliary/ security, red = key telephone). navepoint.com — TIA-606 color codes
- [10]FS.com — Decoding Ethernet Cable Colors: What Do They Mean? States plainly that there is no universal standard dictating copper Ethernet jacket colors, that the cable's performance is identical regardless of color, and documents one common (vendor) convention set for blue, red, yellow, green, orange, purple, white, grey and black. fs.com — Decoding Ethernet Cable Colors
- [11]Turn-key Technologies — Structured Cabling Color Coding Standards. Enterprise by-service convention (including black and pink), the point that color coding is not a technical requirement so “consistency and documentation are more important,” that color without labels leaves the system exposed to human error, and that a documented scheme survives vendor and staff turnover. turn-keytechnologies.com — Color Coding Standards
- [12]Wikipedia — RED/BLACK concept, and CNSSAM TEMPEST/1-13, RED/BLACK Installation Guidance (US Committee on National Security Systems). RED = systems and cabling carrying unencrypted classified (plaintext) signals; BLACK = encrypted or unclassified; the two must be separated and/or shielded to prevent compromising emanations, administered via a Certified TEMPEST Technical Authority. en.wikipedia.org — RED/BLACK concept
- [13]The Fiber Optic Association — Fiber Optic Cable Color Codes (TIA-598). Jacket-color-to-fibre-type table for non-military use: yellow = single-mode (OS1/OS2), orange = OM1/OM2 multimode, aqua = OM3/OM4, lime green = OM5, with the caveat that other colors are permitted when the jacket print identifies the type. thefoa.org — Fiber Optic Color Codes
- [14]CablesAndKits Learning Center — Ethernet Cable Colors Explained. Source for the “color ≠ category” point: jacket color does not indicate category, shielding or gauge; a blue Cat 5e and a blue Cat 6A look identical; the printed jacket markings (and a certification test) are what prove category; and a color scheme should always be documented because no single one is universal. cablesandkits.com — Ethernet Cable Colors