Ethernet vs Wi‑Fi

Ethernet is usually the most predictable choice for latency and sustained throughput; Wi‑Fi wins on mobility. This 2026 guide compares the limits that actually matter in a UK home network.

Connection comparison guide

Choose the right connection for speed, stability and convenience

Ethernet gives you a direct wired link to the router. Wi‑Fi gives you flexibility and freedom around the home. Both have their place, but they behave differently when you test speed, play games, stream video or work from home.

Router with a wired Ethernet connection on one side and multiple wireless devices on the other, representing Ethernet versus Wi‑Fi

2026 quick verdict

Ethernet vs Wi‑Fi: which should you use?

Choose Ethernet if…

You care about repeatable latency, low jitter, large local file transfers or a clean broadband speed-test baseline. A healthy 1GbE path commonly tops out around 940 Mbps of usable internet throughput; 2.5GbE or faster ports are needed to exceed that ceiling.

Choose Wi‑Fi 6E / Wi‑Fi 7 if…

You need mobility and strong multi-device performance. Wi‑Fi 7 can use 320 MHz channels, 4096‑QAM and Multi‑Link Operation, but those are PHY capabilities rather than a guarantee that one phone or laptop will see the headline aggregate data rate.

Best setup for most homes: wire fixed, latency-sensitive devices and use modern Wi‑Fi for phones, tablets and portable laptops.

Baseline test

Use Ethernet to prove whether Wi‑Fi is the problem

An Ethernet test is the cleanest way to separate a broadband fault from a home Wi‑Fi issue. Connect a laptop or desktop directly to the router using a good Cat5e or Cat6 cable, then compare that result with Wi‑Fi in the same room and in the problem room.

If you normally use a VPN, disconnect it for the first baseline. Then reconnect the same VPN server and repeat the test so you can separate home-network performance from VPN routing and encryption overhead. See how a VPN affects speed, ping and routing.

Check the cable as well: read the writing printed on the Ethernet lead. A proper Cat5e, Cat6 or better cable should use all eight internal wires. Some old or very cheap router leads only use four wires, which can cap a speed test at 100 Mbps and make a 300 Mbps, 500 Mbps or gigabit full-fibre line look falsely slow.

5-minute Ethernet baseline: isolate the fault

  1. Pause VPNs and heavy traffic. Disconnect the VPN for the first run and pause large downloads, uploads or cloud backups that could distort the baseline.
  2. Connect directly to the router. Use a known-good Cat5e or better cable and a LAN port fast enough for your broadband package.
  3. Check the negotiated link speed. Confirm the computer has connected at 1Gbps, 2.5Gbps or the expected rate rather than falling back to 100Mbps.
  4. Run the wired test. Record download, upload, idle latency and loaded latency so the wired result becomes your control sample.
  5. Repeat over Wi‑Fi in the problem room. Use the same device and test method. A large deterioration only on Wi‑Fi points to local coverage, interference or client limits rather than the broadband line alone.

Ethernet fast, Wi‑Fi slow

Your broadband line is likely fine. Focus on router position, mesh, Wi‑Fi standards, walls or device limits.

Ethernet also slow

Check the broadband package, provider status, ONT/router port speed, line fault or local congestion before buying Wi‑Fi kit.

Best mixed setup

Wire fixed devices such as consoles, desktops and work docks, then use Wi‑Fi for phones, tablets and flexible rooms.

What is the difference between Ethernet and Wi‑Fi?

Ethernet is a point-to-point wired link between your device and a router or network switch. Modern switched Ethernet is normally full duplex, so a device can transmit and receive on its link at the same time. Common home-network port speeds are 1GbE, 2.5GbE and, on higher-end equipment, 5GbE or 10GbE.

Wi‑Fi carries 802.11 frames over shared radio spectrum. Devices contend for airtime, and the usable data rate changes with signal quality, channel width, interference, retransmissions, client antenna design and the capabilities of both the access point and the client. That variability is why a Wi‑Fi speed test can change substantially without the broadband line itself changing.

Side-by-side Ethernet versus Wi-Fi diagram comparing the direct wired path with shared wireless airtime, mobility, latency and common use cases.
Use Ethernet when consistency matters most; use Wi‑Fi when mobility matters and the radio path is strong.

Which is faster: Ethernet or Wi‑Fi?

There is no useful answer based only on the headline standard name. A 2.5GbE laptop connected to a 2.5GbE router can be faster than a typical Wi‑Fi client, while a modern Wi‑Fi 7 laptop can easily outperform an old 100Mbps or 1GbE wired interface in the right conditions. The real comparison is the slowest link in the complete path: broadband service, router WAN port, router LAN port, Ethernet adapter, cable, Wi‑Fi radio, channel conditions and the client device.

Multi-gigabit bottleneck: a 1GbE port cannot deliver a 1.2Gbps or 2Gbps FTTP service to one wired client at full rate. In practice, a 1GbE internet speed test often lands around 930–950Mbps after Ethernet, IP and transport overhead. To go beyond that, the router and client need 2.5GbE or faster ports. Cat5e can support 2.5GBASE‑T and 5GBASE‑T to 100m when the installed channel meets the standard; Cat6a is the straightforward choice for 10GBASE‑T to 100m.

Wi‑Fi 7 vs Wi‑Fi 6E: why the wireless ceiling moved

Wi‑Fi 6E extends Wi‑Fi 6 (802.11ax) into the 6GHz band. That gives compatible devices access to cleaner spectrum with fewer legacy clients, but the shorter-wavelength 6GHz signal generally has less margin through walls than lower-frequency bands and still depends heavily on distance and line of sight.

Wi‑Fi 7 (802.11be) raises the PHY ceiling with features including channels up to 320MHz, 4096‑QAM and Multi‑Link Operation (MLO). A wider channel can carry more data, while 4096‑QAM encodes more bits per symbol when the signal-to-noise ratio is excellent. MLO can let compatible devices use more than one link across available bands/channels, improving throughput, resilience or latency depending on the implementation. The often-quoted figure of roughly 46Gbps is an aggregate theoretical PHY maximum across many spatial streams, not a realistic speed for one ordinary laptop.

Ethernet vs Wi‑Fi technical comparison for a modern home network
CriterionEthernetWi‑Fi 6E / Wi‑Fi 7
Connection typeSwitched wired link over twisted-pair copper; typically RJ45 using Cat5e, Cat6 or Cat6a.Shared 802.11 radio link between client and access point.
Common home link rates1, 2.5, 5 and 10Gbps depending on ports and cabling.Client PHY rate varies by Wi‑Fi generation, channel width, spatial streams and signal quality.
High-end standard ceiling10GBASE‑T: 10Gbps to 100m on compliant Cat6a. Cat6 can support 10GbE over shorter channels under suitable conditions.Wi‑Fi 7 theoretical aggregate PHY rate is about 46Gbps across the full standard configuration; a single consumer client is far lower.
Usable throughputPredictable but below raw link rate after protocol overhead; about 940Mbps is common on a healthy 1GbE internet path.Highly variable. Multi-gigabit throughput is possible at short range with suitable Wi‑Fi 7 hardware, but distance, contention and client capability reduce it.
Local-link latencyUsually sub-millisecond to the local router/switch on a healthy LAN, with very low variation.Can be only a few milliseconds at close range, but airtime contention, retries and power-saving behaviour can create larger spikes.
Jitter & packet deliveryA healthy wired LAN should have negligible loss and very low jitter; faults usually point to cabling, ports or congestion.Retries can hide RF errors from applications but add delay. Weak signal, busy channels and interference can increase jitter or visible packet loss.
Medium / spectrumBalanced baseband signalling over dedicated copper pairs.2.4GHz, 5GHz and, on 6E/7 equipment, 6GHz spectrum subject to regional rules.
Duplex / contentionModern switched links normally operate full duplex with dedicated send/receive capacity.A channel is a shared contention medium. Conventional operation schedules access to airtime rather than providing each client a dedicated full-duplex wire.
Best useCompetitive gaming, NAS, workstation docks, local backups, media editing and broadband fault isolation.Phones, tablets, portable laptops, roaming clients, smart-home devices and rooms where cabling is impractical.

Why does Wi‑Fi slow down through walls?

Wireless performance is governed by the link budget: every metre of distance and every obstacle removes signal margin. Dense masonry and reinforced concrete can absorb and scatter radio energy; metal-backed plasterboard or foil insulation can act as a partial RF barrier; low-emissivity glazing may contain metallic coatings that attenuate signals. The result is usually a lower modulation rate, more retransmissions and less usable throughput rather than an abrupt on/off failure.

Higher frequencies make the trade-off more obvious. The 6GHz band gives Wi‑Fi 6E and Wi‑Fi 7 access to wide, relatively clean channels, but it is best used where the access point is nearby. For difficult rooms, moving the access point, adding a correctly placed mesh node or using wired backhaul often helps more than buying a faster broadband package. See the UK building-materials and Wi‑Fi attenuation guide.

Interactive Wi‑Fi planning tool

Wi‑Fi wall & obstacle impact estimator

Choose the main obstacle and Wi‑Fi band to estimate how likely the path is to reduce signal margin, force a lower modulation rate or increase retries. This is a planning guide, not a substitute for an RF survey.

Estimated impact: Low to moderate

Plain plasterboard usually leaves enough margin for a healthy 5GHz connection when the access point is nearby, although distance and client hardware still matter.

Ethernet control: once a compliant cable is installed through or around the obstacle, the wall itself does not consume radio signal margin. Link speed is then governed by the Ethernet ports, cable category/quality and run length.

Why there is no fake “% speed loss” figure: wall loss does not translate cleanly into a fixed throughput percentage. Thickness, moisture, reinforcement, metallic coatings, angle, distance, access-point power, antenna design and neighbouring networks can all change the result.

Why Ethernet is usually better for gaming

Online games rarely need enormous download bandwidth. What matters is latency consistency. Jitter is variation in packet delay; packet loss is data that never reaches its destination and must be recovered or tolerated by the application; bufferbloat is excessive queueing delay when a bottleneck becomes busy. These are different problems, and bufferbloat can affect Ethernet or Wi‑Fi because the queue may sit in the router, modem or upstream network.

Ethernet removes the wireless airtime variable. A switched full-duplex Ethernet link gives the console or PC a dedicated local connection, while Wi‑Fi devices share airtime and may wait, retry or change modulation as RF conditions change. Wi‑Fi 7 MLO can improve wireless latency and reliability, but it does not make the radio environment identical to a cable. For competitive gaming, a good Ethernet path remains the simplest way to remove one source of jitter.

Test the right thing: if gaming lags while download speed looks healthy, run the bufferbloat test, check loaded latency, then repeat over Ethernet. A large improvement on the wired run points toward the local Wi‑Fi path rather than the broadband package alone.

Which is better for streaming and everyday use?

For a phone, tablet or streaming stick, good Wi‑Fi is normally the sensible choice. A stable 4K stream uses far less bandwidth than a modern Wi‑Fi link can carry. Ethernet becomes more valuable when a fixed TV, media server or console is in a weak-signal location, or when you want predictable performance while other wireless devices are busy.

Which is better for working from home?

For video calls, remote desktops, VPN sessions and large file transfers, stability is usually more valuable than peak speed. A wired dock can remove roaming, weak-signal and channel-contention issues from a fixed desk. Wi‑Fi remains the better choice when mobility matters, provided the work area has strong coverage and the access point is not overloaded.

Use Ethernet for…

Gaming PCs and consoles, desktops, NAS devices, fixed workstations, high-volume local transfers and accurate broadband baseline tests.

Use Wi‑Fi for…

Phones, tablets, portable laptops, smart-home devices and everyday clients that move around the property.

Best hybrid setup

Use Ethernet as the network backbone and for fixed critical devices, then let Wi‑Fi handle mobility. Wired backhaul can also improve mesh performance.

When should you test over Ethernet?

If you are diagnosing slow broadband, Ethernet is the control sample. It answers a crucial question: is the internet connection slow, or is the local wireless path slow?

  • Use a known-good Cat5e or better cable and a device whose Ethernet port is fast enough for the package.
  • Check the negotiated link speed in the operating system; a 100Mbps negotiation can make a fast FTTP service look faulty.
  • If Ethernet reaches the expected service speed but Wi‑Fi does not, troubleshoot coverage, channel use, router position and client capability.
  • If Ethernet is also slow, check the ONT/router WAN and LAN port rates, provider service, local congestion and the test device before buying new Wi‑Fi hardware.

Interactive decision tool

Should I cable it?

Choose your main activity and broadband tier. The recommendation focuses on the connection path, not on selling you a faster package.

How to choose between Ethernet and Wi‑Fi

  • Choose Ethernet when latency consistency, local file speed or troubleshooting accuracy matters more than mobility.
  • Choose Wi‑Fi when the device moves around and the room has strong coverage from a modern access point.
  • Do not buy Cat6a just because your internet exceeds 1Gbps. For 2.5GbE, existing compliant Cat5e often works. Cat6a becomes especially useful when you are planning 10GbE to 100m or want generous cabling margin.
  • If you rent or cannot run new cable, consider a well-positioned mesh system with wired backhaul where possible. Powerline can work but varies with electrical wiring and noise; usable coaxial cabling can also provide a wired alternative with suitable adapters.

Networking glossary

Core Ethernet and Wi‑Fi terms explained

These terms explain why the headline speed printed on a router, cable or Wi‑Fi standard is not the same as the performance a device actually experiences.

PHY rate / PHY ceiling
The raw physical-layer signalling rate of a network link. It is measured before protocol overhead, retransmissions, contention and application-level limits, so usable throughput is always lower than the headline PHY figure.
Full duplex
A modern switched Ethernet link can transmit and receive at the same time on its dedicated connection. This removes the shared-radio airtime contention that wireless clients must manage.
Wi‑Fi airtime contention
Wi‑Fi devices share radio spectrum and coordinate access to the channel. Busy neighbouring networks, retries and many active clients can increase waiting time even when the broadband line itself is fast.
Jitter
Variation in packet delay over time. A connection can have a reasonable average ping but still feel poor for gaming, calls or remote desktop work if individual packets arrive with uneven timing.
Bufferbloat
Excessive queueing delay when a network bottleneck becomes busy. Large router or modem queues can make latency jump during downloads or uploads even on a wired Ethernet connection.

Frequently asked questions

Is Ethernet always faster than Wi‑Fi?

No. Speed depends on the link at both ends. A modern Wi‑Fi 7 client can outperform an old 100Mbps or 1GbE wired interface, but a correctly designed 2.5GbE, 5GbE or 10GbE Ethernet path is more predictable and is not affected by radio distance or neighbouring Wi‑Fi traffic.

Why is Ethernet usually better for online gaming?

Ethernet removes wireless airtime contention, signal changes and RF retransmissions from the local path. That normally gives more consistent latency and jitter. It does not cure internet-side congestion or bufferbloat elsewhere in the route.

Can Wi‑Fi 7 replace an Ethernet cable?

For many phones and laptops, yes. Wi‑Fi 7 can provide very high throughput and lower latency with features such as MLO. Ethernet still has the advantage for fixed devices where predictable latency, local transfers or fault isolation matter most.

Do I need Cat6a for multi-gigabit broadband?

Not automatically. 2.5GBASE‑T and 5GBASE‑T were designed to work over compliant Cat5e/Cat6 channels up to 100m. Cat6a is the standard choice for 10GBASE‑T to 100m and is a sensible option for new permanent cabling when you want 10GbE headroom.

Why does a 1Gbps Ethernet speed test often show about 940Mbps?

A 1GbE link rate includes framing and protocol overhead, so the application payload is lower than 1000Mbps. A result around the low-to-mid 900Mbps range can therefore be normal on a healthy gigabit wired path.

Should I run a broadband speed test over Wi‑Fi or Ethernet?

Use Ethernet first when diagnosing the broadband line. Check that the device negotiated the expected port speed, then compare Wi‑Fi in the same room and in the problem room. That separates WAN performance from wireless coverage and interference.

How LinkSpeed tests and checks claims