Within its supported distance, a compliant Ethernet link does not become a little slower for every extra meter of cable. A healthy 5 m and 50 m Cat5e or Cat6 link can both negotiate at 1 Gbps. The familiar 100 m (328 ft) limit is a standards-based channel boundary, not a cliff where the signal disappears at 100.1 m. Beyond that boundary, a link may still come up, but error rate, link stability, and Power over Ethernet performance are no longer assured by the normal channel specification. If a 20 m replacement cord suddenly links at only 100 Mbps, suspect the cord, plug, port, or termination before assuming that 20 m is inherently too long.
- Ethernet link rates are negotiated in defined steps; they do not normally decline in proportion to cable length.
- The usual 100 m limit covers the complete copper channel between active devices, not just the cable hidden in the wall.
- A common structured-cabling model uses up to 90 m of permanent link plus up to 10 m of patch and equipment cords in total.
- A link beyond 100 m may work, but it sits outside the normal standards assurance for transmission and PoE performance.
- For longer routes, use an accessible switch, fiber, or a manufacturer-validated extended-reach system instead of relying on a higher Category number.
1. Does a longer Ethernet cable reduce speed?
Usually not within the supported channel length. Ethernet devices do not measure the cable and subtract bandwidth for every foot or meter. The two endpoints advertise and negotiate supported modes such as 100 Mbps, 1 Gbps, 2.5 Gbps, or 10 Gbps. If the complete cabling path meets the requirements for the selected application, the link operates at that defined rate.
This is why replacing a 2 m patch cord with a sound 20 m patch cord should not make a normal gigabit link negotiate at 800 Mbps. Both links can show 1.0 Gbps. An Internet speed test may still vary because of the ISP, test server, router, storage, CPU load, VPN, or other traffic, but that variation is not proof that the extra 18 m linearly reduced Ethernet speed.
Length does have physical effects. Signal loss increases with distance, and the time for a signal to travel through the cable also increases. Cabling and Ethernet specifications account for those effects within a defined operating envelope. At home or in a small office, the propagation-time difference between a short and a tens-of-meters cable is tiny compared with the delay across an ISP and a remote server. For gaming, server location, routing, congestion, Wi-Fi, and buffer behavior matter far more when both wired links are healthy.
| What you observe | Likely interpretation | First check |
|---|---|---|
| A 2 m and 20 m cord both link at 1 Gbps | No reason to expect a linear speed penalty from length alone | Compare throughput under the same wired test conditions |
| The longer replacement links at 100 Mbps | The cord, plug, port, or pair continuity may be faulty | Use a known-good cord and inspect the negotiated link rate |
| Link remains 1 Gbps but speed tests fluctuate | The local link is gigabit; another performance variable is changing | Test without Wi-Fi, VPN, downloads, or other users |
| A run near 100 m has errors or intermittent drops | Channel margin, termination, cable quality, temperature, or length may be involved | Verify the whole route and run the correct certification test |
2. What does the 100-meter Ethernet limit include?
The often-quoted 100 m limit normally refers to the entire balanced-copper channel between two active Ethernet interfaces. In a commercial installation, that path can include a switch equipment cord, patch-panel connection, horizontal cable, work-area outlet, and user equipment cord. Measuring only the visible cable at the desk misses most of the channel.
A common TIA and ISO/IEC structured-cabling model allows a permanent link of up to 90 m (295 ft) and patch or equipment cords totaling up to 10 m (33 ft), producing a 100 m channel. A 5 m + 90 m + 5 m drawing is an easy example, but it does not mean each end automatically receives 10 m. The total cord allowance is shared, and specific designs can require derating for cord construction, conductor size, temperature, connections, or application conditions.
cord
up to 90 m
cord
Typical model: permanent link up to 90 m, with patch and equipment cords bringing the complete channel to no more than 100 m. Confirm the adopted standard and system conditions for the actual project.
In a home, one factory-terminated cable may connect the router directly to a computer. In that case, the cord is effectively the whole copper channel. In an office, school, warehouse, or data center, the route is usually segmented. Include rack cords, patch panels, consolidation points, outlets, couplers, and device cords when evaluating distance and performance.
3. What happens when an Ethernet run exceeds 100 m?
Nothing magical happens at precisely 100.1 m. The 100 m value is a design boundary intended to provide interoperable performance under the standard’s specified worst-case conditions. A 105 m link may work because its components and environment have extra margin. A poorly installed 70 m link can fail because the cable is damaged, connectors are badly terminated, or noise and temperature have consumed that margin.
As distance increases, insertion loss grows. More connectors, excessive untwisting, sharp bends, crushing, poor impedance matching, or questionable conductor materials can add loss and reflections. The receiver may then see frame errors, retransmissions, intermittent link drops, failure to negotiate the intended rate, or no link at all. A basic web page or a lit port LED does not demonstrate low error rates under sustained traffic.
Power over Ethernet adds another constraint. Copper conductors have resistance, so some voltage is lost and some power becomes heat. Longer resistance paths, smaller conductors, higher current, large cable bundles, and high ambient temperature can reduce the power margin available to the powered device. A camera or access point may start normally and then reboot when heaters, radios, motors, or infrared LEDs raise its load.
4. Why can a short cable fail while an overlength cable works?
Length is only one part of channel performance. A short patch cord can fail because a modular-plug contact is not seated, one twisted pair is open, the cable has been crushed, or the conductors and construction do not meet the claimed specification. A gigabit link uses all four pairs. If a required pair is unavailable, two gigabit-capable devices may establish only a 100 Mbps link or fail to link, depending on the fault and the endpoints.
An overlength link can work today because the actual components outperform their minimum specifications and the installation has favorable noise, temperature, and connection conditions. That does not make the design standards-compliant. A different patch cord, warmer ceiling space, higher PoE load, added connection, or replacement switch could remove the remaining margin.
Cable-test length results also need context. Field certifiers estimate electrical length from signal travel time and the cable’s nominal velocity of propagation (NVP). Different pair twist rates can produce slightly different electrical lengths, and an incorrect NVP setting changes the displayed result. Test standards may account for measurement uncertainty, but that allowance is not a license to design an intentionally longer permanent link. Other parameters can still fail even if the displayed length receives a PASS.
| Common claim | More accurate interpretation |
|---|---|
| “Ethernet loses a fixed percentage of speed every 10 m.” | Healthy links negotiate defined rates; distance does not normally create a smooth proportional reduction. |
| “The signal dies the instant the channel exceeds 100 m.” | 100 m is a standards assurance boundary, not a precise physical cliff. |
| “My 105 m link works, so the installation passes.” | Current connectivity does not prove standards compliance or long-term margin. |
| “Cat7 or Cat8 automatically removes the distance limit.” | Application, channel topology, components, and the governing specification determine reach. |
| “The tester passed length, so every requirement passed.” | Length is only one result; wire map, loss, return loss, crosstalk, and other limits still matter. |
5. How cable category, speed, and distance fit together
Do not ask only, “How far can Cat6 go?” The complete question is, “Which Ethernet application must this installed channel support, using which cabling system and topology?” The familiar 100 m reach applies to many—but not every—twisted-pair Ethernet combination.
Category 5e and Category 6 are both established choices for 1000BASE-T over a compliant 100 m channel. Category 6A is designed to support 10GBASE-T over a 100 m channel. Category 6 can support 10GBASE-T over shorter distances under applicable installation and alien-crosstalk conditions, but it should not be described as universally providing 10 Gb/s over 100 m.
Category 8 shows why a larger Category number does not automatically mean longer reach. Category 8 was developed for 25GBASE-T and 40GBASE-T in a two-connector channel up to 30 m (98 ft), mainly for data-center switch-to-server links. Its performance is higher, but the target applications use a shorter channel. Select category and distance together with the required Ethernet application.
| Typical application | Common cabling choice | Standards-oriented channel reach | Important qualification |
|---|---|---|---|
| 1000BASE-T (1 Gb/s) | Cat5e or better | Up to 100 m | Complete four-pair channel must meet the required performance |
| 10GBASE-T (10 Gb/s) | Cat6A | Up to 100 m | Cat6 reach is installation-dependent and shorter |
| 25GBASE-T / 40GBASE-T | Cat8 | Up to 30 m, two connectors | Data-center application; not a method for extending ordinary LAN reach |
For a home gigabit connection under a few tens of meters, a trustworthy, correctly terminated Cat5e or Cat6 cord is normally sufficient. Pure-copper construction, reliable plugs, realistic product specifications, and the actual negotiated link rate matter more than a marketplace listing that simply advertises Cat7 or Cat8.
6. How to estimate and test the actual cable length
For a packaged patch cord, start with the label, order record, and jacket marking. Some bulk cables have sequential meter or foot markings, so installers can estimate deployed length from the difference between the start and end values. Include slack and service loops; do not count only the straight-line distance between rooms.
For concealed cabling, floor-plan distance is often too optimistic. Add rack routing, vertical rises, ceiling-tray detours, wall drops, and service loops. Then include cords at both ends and every intermediate connection. In a commercial project, the pathway drawing and cable schedule should agree with field labels and test records.
A field certification tester estimates length using propagation time and a configured NVP. Select the correct cable type or enter the manufacturer’s NVP when required. Pair-length differences can be normal because the four pairs use different twist rates. Review the full Autotest for the selected Category and permanent-link or channel model; a standalone length reading does not certify transmission performance.
7. What should you use beyond 100 meters?
The simplest standards-based option is often an Ethernet switch at an accessible intermediate location, creating two separate copper links within their permitted reach. The switch needs reliable power, ventilation, physical security, and maintenance access. Hiding an unmanaged switch above a ceiling may solve distance today while creating a difficult failure point later.
Fiber is usually the stronger choice for building-to-building links, campus backbones, long indoor routes, high bandwidth, electromagnetic-interference exposure, or locations with grounding-potential differences. Select the fiber type, optics, connectors, and loss budget as a system. Fiber does not carry ordinary PoE, so power at the remote endpoint must be planned separately.
PoE extenders and vendor-specific extended-reach copper systems can fit cameras, sensors, or other fixed applications. They are not interchangeable with a generic 100 m channel. Verify the exact supported speed, cable type, number of intermediate devices, output power at the endpoint, temperature rating, ingress protection, and permitted topology in the manufacturer’s documentation. Each extender also consumes power and becomes another point to maintain.
- Only a few meters over: remove needless detours or relocate the telecommunications outlet, while preserving a sensible service loop.
- Accessible indoor midpoint: use a properly powered and managed switch so each link stays within reach.
- Long backbone or building crossing: use an engineered fiber link.
- Remote PoE camera or sensor: evaluate a documented extender or extended-reach system, including endpoint power.
- Commercial acceptance: test the exact installed topology against the specified limit; do not approve it from a link LED alone.
8. Frequently asked questions
Will a 30 m Ethernet cable increase gaming ping?
Not by a meaningful amount when both the short and long cables are healthy and negotiate the same rate. Internet routing, server distance, congestion, router queues, and Wi-Fi add far more delay. A damaged longer cord can create errors or instability, but that is a fault, not the normal effect of 30 m.
Does Ethernet stop working after 100 m?
Not necessarily. It may link beyond 100 m, but the normal 100 m channel specification no longer assures performance. Do not base a new installation on an overlength link that happened to work during a brief test.
Can Cat7 or Cat8 run farther than Cat6?
Not as a general rule. Reach depends on the Ethernet application and channel specification, not the marketing number alone. Cat8’s headline 25/40 Gb/s applications use a 30 m channel. For ordinary long-distance LAN links, fiber or an intermediate switch is usually clearer.
Can I join cables with couplers and stay under 100 m?
A coupler becomes part of the channel and adds another connection whose performance must be suitable for the system. One compliant consolidation point may be part of a designed topology. A chain of unverified couplers is not equivalent to a properly installed and tested channel, even when the total measured length is under 100 m.
Is a PoE extender enough for a CCTV camera beyond 100 m?
Only if the complete, documented combination supports the required distance, data rate, endpoint power, temperature, and environment. Check the camera’s maximum load, extender consumption, cable requirements, and the manufacturer’s topology limits.
9. Conclusion: design a healthy channel, not just a cable length
Ethernet cable is not a water pipe that loses a fixed percentage of speed with every extra meter. Within the supported channel distance, compliant components and correct installation should maintain the intended negotiated rate. The 100 m boundary defines the normal end-to-end copper channel, commonly built from a 90 m permanent link plus no more than 10 m of cords in total.
For a home or small office, a sound 20 m or 30 m Cat5e or Cat6 cable is not inherently slow. If performance changes, check the link rate, cable, connectors, ports, and test conditions. For a commercial route near or beyond 100 m, calculate the entire channel and choose a deliberate architecture—an accessible switch, fiber, or a validated extended-reach system—then test the installation against the correct limit.
Read next
Sources
- Telecommunications Industry Association — ANSI/TIA-568.2-E announcement
- ISO/IEC JTC 1/SC 25 — Interconnection of information technology equipment
- IEEE Standards Association — IEEE 802.3 Ethernet
- Fluke Networks — Top 10 Cable Testing Mistakes to Avoid
- Fluke Networks — Extending and Testing Cable Runs Beyond 100 Meters
- Fluke Networks — Channel, Permanent Link, Patch Cords, MPTL, E2E… Oh My!
- Fluke Networks — Category 8 Cabling Fact Sheet
Korean version: 한국어로 읽기
