Ethernet vs Wi-Fi: The Jitter Difference Is Bigger Than You Think

Everyone knows Ethernet is "better" than Wi-Fi. But better by how much, exactly? We ran controlled jitter tests across multiple scenarios — idle network, moderate load, heavy background traffic — on both connection types under identical conditions. The magnitude of the difference surprised us, and we think it'll surprise you too.

Test Methodology

Before the data, the method. We ran all tests using jitter.is against Cloudflare's global edge network, which routes to the nearest PoP and provides consistent, low-variance baseline server-side conditions. Each test session ran for 60 seconds, collecting approximately 50–60 ping samples, with warm-up pings discarded as per jitter.is's standard methodology.

Three load scenarios were tested for each connection type. Idle: only the test device active, no background traffic. Moderate load: one 4K Netflix stream running simultaneously on a second device. Heavy load: two simultaneous large file downloads saturating approximately 85% of available download bandwidth.

All Ethernet and Wi-Fi tests were run within 10 minutes of each other to control for time-of-day ISP variance. The same browser session was used for all tests to eliminate client-side variation.

Results: Idle Network

The first surprise was the idle results. Even with no other devices or traffic, Wi-Fi jitter was already noticeably higher than Ethernet — before any load or contention was introduced.

Even at idle — 4m line-of-sight from a modern Wi-Fi 6 router with no other devices active — Wi-Fi jitter was 14× higher than Ethernet. The 22ms spike visible in the Wi-Fi data is characteristic of the CSMA/CA backoff mechanism: even with no competing devices, the radio protocol introduces occasional variable delays as it checks the channel before transmitting.

The Ethernet result of 0.3ms represents essentially the floor — near-zero jitter on a dedicated full-duplex link with no contention whatsoever. Anything under 1ms on Ethernet is effectively perfect for all real-time applications.

Results Under Load

The data reveals two important patterns. First, Ethernet jitter is nearly invariant to load. It goes from 0.3ms at idle to 1.7ms under heavy simultaneous downloads — a small absolute change that stays well within "excellent" territory throughout. Second, Wi-Fi jitter is extremely sensitive to load. A single 4K stream on another device pushed Wi-Fi jitter from 4.2ms to 18.4ms. Two simultaneous downloads pushed it to 47.8ms — into "poor" territory where gaming and VoIP applications show clear degradation.

Why the Gap Is So Large

The numbers reflect fundamental physics and protocol design, not incidental implementation differences.

Ethernet Is a Private Pipe

A Cat6 cable between your device and router is a dedicated, full-duplex link. Your packets share that wire with no one. There is no contention, no waiting, no checking whether someone else is transmitting. Your packet goes in one end and comes out the other at a near-fixed delay determined only by the speed of electrical propagation and switch processing — both of which are extremely consistent. The variance in delivery time (jitter) approaches zero.

Wi-Fi Is a Shared Radio with Turn-Taking

Every device on your Wi-Fi network — including devices on neighboring networks using the same channel — shares a single half-duplex radio medium. Only one device can transmit at a time. Before transmitting, every device listens to check if the channel is clear, and waits a random backoff period if it isn't. This randomness is structural and unavoidable — it is literally mandated by the 802.11 protocol (CSMA/CA) to prevent collisions. The waiting time before each packet gets to transmit varies unpredictably based on how many devices are competing and what they're doing. That variation is jitter.

Load Amplifies Wi-Fi's Weakness

When background traffic increases, more packets from more devices compete for the shared radio medium simultaneously. Contention increases, backoff intervals grow, and the variance in transmission timing increases proportionally. This is why Wi-Fi jitter scales badly with load — every additional active device on the network makes the problem worse. Ethernet is immune to this: each device has its own wire, and switch architectures prevent contention between separate ports.

What These Numbers Mean in Practice

The most significant finding is the isolation property of Ethernet. On Wi-Fi, what other people in your house are doing directly affects your gaming and call quality. On Ethernet, it doesn't. Someone downloading a game update on a different device cannot cause your jitter to spike. This isolation is arguably as valuable as the raw jitter number itself.

How Much Does Wi-Fi Improve with Better Hardware?

We ran an additional comparison to test whether Wi-Fi 6E's 6GHz band — with its dramatically less congested spectrum — closes the gap with Ethernet in a typical environment.

Wi-Fi 6E's 6GHz band delivers a meaningful improvement — 6.9ms average jitter under moderate load vs 18ms for 5GHz. The 6GHz band's near-empty spectrum in most residential environments dramatically reduces the contention-related variance that dominates 2.4GHz and 5GHz performance. But it still trails Ethernet by a factor of 11× at moderate load. Wi-Fi 6 (5GHz) vs Wi-Fi 5 (5GHz) shows essentially no improvement — the bottleneck is the shared medium and contention protocol, not the generation's throughput improvements.

When You Truly Can't Run a Cable

Ethernet is the right answer for any device where connection quality matters. But there are real situations where running a cable isn't practical. In that order, here are the alternatives:

Summary

The test data leaves no room for ambiguity. Under identical conditions to the same target, Ethernet delivered 0.3–1.7ms jitter across all load scenarios. Wi-Fi delivered 4.2–47.8ms. The gap widens dramatically as household network activity increases — the exact conditions under which you most need stable latency.

The critical insight is that Ethernet's jitter is nearly load-invariant. It doesn't care what else is happening on your network. Wi-Fi's jitter is acutely load-sensitive: every device streaming, downloading, or syncing in the background adds contention to the shared radio medium and pushes your jitter higher.

If your device is within cabling reach of your router, a Cat6 cable is the single most cost-effective networking improvement available. It costs under $10, takes five minutes to install, and delivers a 10–80× jitter reduction depending on your household's background traffic. If direct cable routing isn't feasible, MoCA adapters are the next best option and deliver near-Ethernet performance in homes with coaxial wiring.

Wi-Fi 6E is a genuine improvement and worthwhile if a cable truly isn't possible — but it still can't match the isolation property that makes Ethernet so decisive: on a wired connection, other people's network usage simply doesn't affect your jitter.