Your speed test says 500 Mbps. Your boss still sounds like a robot on the 9 a.m. Teams call, and your squad says you're teleporting again. Download speed was never the problem. The number that wrecked both is usually smaller than 50, it's measured in milliseconds, and most people scroll right past it.
Our take: paying for a faster plan is the classic trap. Jitter usually starts inside your home, on Wi-Fi or in an overloaded router queue, and neither one gets better when you buy more megabits. A free two-ping test and a cheap Ethernet cable fix more calls than any upgrade. Your ISP is guilty less often than you think, and the test further down settles it either way.
What is jitter, and how is it different from ping and packet loss?
Three numbers, three different failures:
- Ping (latency): how long a packet takes to reach a server and come back.
- Jitter: how much that time changes from one packet to the next.
- Packet loss: the share of packets that never come back.
Averages lie. Say your pings come back at 20, 22, 21 and 23 ms. Low jitter, smooth call. Now say they come back at 20, 80, 25 and 95 ms. The average is a respectable 55 ms, and the connection is a mess.
Calls suffer because voice and video apps keep a jitter buffer that holds packets back for a moment to smooth out the gaps. When a packet arrives later than the buffer can wait, the app treats it as lost. That's the robot voice, the clipped word, the frozen face.
The math behind the number matters too. RFC 3550, the RTP standard behind most call apps, defines interarrival jitter as a running average that moves only 1/16 of the way toward each new difference between packets. One spike barely moves it. Steady wobble drives it up.
Your path to the internet runs from your laptop, over Wi-Fi, to your router, then your modem, then your ISP's network, and finally to a far server such as Google Public DNS at 8.8.8.8. A ping to the router covers only the first short stretch: the Wi-Fi or cable between you and the box. A ping to 8.8.8.8 covers the whole path. Hold that thought. The test further down is built on it.
So how much wobble is too much?
What jitter is acceptable for Zoom, Microsoft Teams and online gaming?
Microsoft is the strictest. As of November 2026, Teams guidance asks for jitter under about 30 ms and packet loss under about 1% over a 15-second window, with latency around 50 ms one way. As of November 2026, from a company's network edge to Microsoft's, the guidance is tighter, around 15 ms of jitter.
Zoom is looser. As of November 2026, Zoom's guidance asks for roughly 150 ms latency or less, jitter of about 40 ms or less and packet loss under about 2%.
Games have no official number. As of November 2026, we found no jitter limit published by Valve, Riot, Epic or Sony. Our rule of thumb for fast shooters is jitter of 20 ms or less, ping of 60 ms or less and loss of 1% or less. Turn-based and slower games forgive far more.
The blunt part: a connection that's "fine for Zoom" can fail Teams. Any test that hands you a single good-or-bad verdict is hiding that from you.
Teams and Zoom network limits side by side, plus our gaming rule
| Source | Measured between | Latency | Jitter | Packet loss | Checked |
|---|---|---|---|---|---|
| Microsoft Teams | Your device to Microsoft network edge | Under 50 ms one way, under 100 ms round trip | Under 30 ms in any 15 s | Under 1% in any 15 s | Nov 2026 |
| Microsoft Teams | Your network edge to Microsoft network edge | Under 30 ms one way, under 60 ms round trip | Under 15 ms in any 15 s | Under 1% in any 15 s | Nov 2026 |
| Zoom | Your device to Zoom | 150 ms or less | 40 ms or less | Under 2% | Nov 2026 |
| ITU-T G.114 (05/2003) | Mouth to ear, one way | About 150 ms acceptable to most users, 400 ms outer limit | No figure set | No figure set | Nov 2026 |
| Online gaming (site rule of thumb, no official limit) | Your device to game server | 60 ms or less | 20 ms or less | 1% or less | Nov 2026 |
Microsoft Learn: Prepare your organization's network for Microsoft Teams; Zoom Support network guidance; ITU-T Recommendation G.114. All checked November 2026. Gaming row is the site's own guideline.
Is 35 ms jitter bad, and at what number do calls and games start breaking up?
35 ms fails Teams, passes Zoom and is too high for competitive gaming by our rule. On Teams, expect the occasional robot syllable. In a shooter, expect rubber-banding.
The standard behind every call limit is ITU-T G.114, "One-way transmission time", in force in its 05/2003 edition. Its Appendix II, guidance on one-way delay for Voice over IP, arrived in Amendment 1 (09/2003) and is included in that edition. G.114 is commonly read as treating about 150 ms one way as the point up to which most users don't notice delay, and 400 ms as the outer limit for network planning.
Jitter eats into that budget. Every millisecond the jitter buffer holds a packet back is added to the delay. High jitter forces a bigger buffer, so the call feels laggy even when your ping looks fine.
Rough breaking points:
- Under 20 ms: most people won't notice, as a rule of thumb as of 2026.
- 30 to 50 ms: borderline by our rule of thumb, depending on the app's buffer.
- Above 50 ms with any packet loss: by our rule of thumb, calls tend to break up and games stutter.
Jitter results checker: which app limits your connection passes
Example: Jitter: 35 ms, Ping (round trip): 45 ms, Packet loss: 0.5 %. Result: 1 of 3.
How do I run a jitter test, and why do different speed tests show different jitter numbers?
The quick options:
- Speedtest by Ookla shows jitter next to ping.
- Cloudflare's speed test at speed.cloudflare.com shows jitter plus latency while the line is loaded.
- Fast.com, under "Show more info", shows unloaded and loaded latency. It gives no jitter figure, but the gap between the two is a quick bufferbloat check.
They disagree for good reasons. Each tool uses its own formula: the average difference between pings, a standard deviation, or an RFC 3550-style smoothed value. Each takes a different number of samples from a different server at a different distance. And some measure an idle line while others measure it under load. A 4 ms reading on one site and 12 ms on another can both be correct.
No speed test tells you where the jitter happens. Two pings do.
How do you ping your router and 8.8.8.8 to see if the jitter is your Wi-Fi or your line?
Five steps, about ten minutes:
- Find your router's address. On Windows,
ipconfigprints a "Default Gateway" line. On a Mac, open Terminal androute -n get defaultprints a "gateway" line. It's often 192.168.1.1 or 10.0.0.1. - Ping the router 100 times on Wi-Fi.
- Ping 8.8.8.8 100 times on Wi-Fi.
- Plug in an Ethernet cable and repeat both.
- Compare the four results with the decision tree below.
Copy-ready commands (swap in your own gateway address):
Windows (Command Prompt)
ipconfig
ping -n 100 192.168.1.1
ping -n 100 8.8.8.8
Mac (Terminal)
route -n get default
ping -c 100 192.168.1.1
ping -c 100 8.8.8.8
Here's a sample of what comes back, with the parts that matter marked:
Mac
100 packets transmitted, 100 packets received, 0.0% packet loss <- packet loss
round-trip min/avg/max/stddev = 2.1/4.8/31.6/4.2 ms <- stddev = your jitter
Windows
Packets: Sent = 100, Received = 99, Lost = 1 (1% loss) <- packet loss
Minimum = 2ms, Maximum = 48ms, Average = 6ms <- jitter = the spread
A Mac prints a standard deviation, which works as a jitter figure. Windows doesn't, so read the gap between Minimum and Maximum, then scroll up for single lines with big times. Linux prints "mdev", which you read like the Mac stddev.
The decision tree:
- Router ping on Wi-Fi: is jitter high? That means a Mac stddev above about 10 ms, or a Windows Maximum dozens of milliseconds above the Minimum. Yes: go to 2. No: go to 3.
- Same router ping on a cable: is it low now? Yes: Wi-Fi problem, so start with the free and cheap Wi-Fi fixes below. No: router or device problem, so restart, update the router firmware, try another device, then consider a new router.
- 8.8.8.8 ping on a cable: high jitter or any loss? Yes: line or ISP problem, so log it at different times and take it to your ISP. No: go to 4.
- 8.8.8.8 ping while a big upload or download runs: does jitter jump? Yes: bufferbloat, so turn on SQM or QoS. No: your network is clean, and the jitter comes from the app's server or your VPN route.
Why is jitter high on Wi-Fi but fine on cable, and why does it spike in the evening?
Wi-Fi is shared airtime. Every device in your home and every neighbor's network on the same channel takes turns, and each retry after interference shows up as jitter. The 2.4 GHz band is the worst in apartment buildings, and microwaves and thick walls make it worse. A cable has no airtime to fight over. If your router ping drops from 40 ms of jitter on Wi-Fi to 3 ms on a cable, case closed.
Evenings are a different story. As of 2026, neighborhood cable nodes and ISP links are typically busiest in the evening, roughly 7 to 11 p.m. If router pings stay flat while 8.8.8.8 pings climb at night, the congestion is upstream of your house.
Then there's bufferbloat. Someone uploads a video, a phone starts a photo backup, a console grabs a 100 GB update. The router's oversized buffer fills, and every packet, your call included, waits in line behind that transfer. Say your idle ping is 15 ms: under load it can hit 200 ms. As of November 2026, Cloudflare's speed test shows this as loaded latency, and the Waveform bufferbloat test grades it on a letter scale.
Here's what catches people out: uploads usually hurt more than downloads. Home upload capacity is far smaller, so on a cable plan of, say, 500 Mbps down and 20 Mbps up, one cloud backup fills the upload pipe.
Bufferbloat in one example: 15 ms at rest, 200 ms during an upload
- 15 msExample idle ping with nothing else running
- 200 msThe same ping while a big upload fills the router's buffer
Does jitter matter more than ping for gaming, and does a VPN make it worse?
Steady beats fast. Game netcode can predict around a constant 50 ms ping. It can't predict a ping that jumps between 20 and 90 ms, so you get rubber-banding, shots that don't register and players who teleport.
Our view: for competitive play, a stable 45 ms beats a jittery 25 ms. Fix jitter first, then chase ping.
A VPN adds a hop, encryption overhead and often a crowded server, so jitter usually rises. The exception is an ISP with a bad route to a game server. There, a VPN takes a different path and can lower both ping and jitter. Run the 8.8.8.8 test with the VPN on, then off, and let the numbers decide.
The most common mistake? Blaming the game server when the jitter already shows up in the router ping.
How do I lower jitter at home, from the cheapest fix to the most effective?
Free:
- Pause cloud backups and big downloads during calls and matches.
- Move closer to the router, or raise it off the floor and away from metal.
- Switch the laptop to the 5 GHz or 6 GHz band.
- Restart the modem and the router.
Cheap:
- An Ethernet cable to the desk or console, which cost roughly $10 to $15 for 25 feet of Cat 6 as of 2026. It's the single most effective fix for Wi-Fi jitter.
- Powerline or MoCA adapters where you can't run a cable.
Router settings:
- Turn on QoS or, better, Smart Queue Management (fq_codel or CAKE), set to about 90% of your measured download and upload speeds. That's the real cure for bufferbloat.
Bigger moves:
- A mesh system or a newer router for a large home.
- A different plan or technology, such as fiber instead of cable or DSL, but only once the two-ping test points at the line.
Re-test after each change, so you know which fix actually worked.
When is high jitter the ISP's fault, and what proof should you bring when you call?
It's the ISP when, on a cable straight to the router, router pings are clean and pings to 8.8.8.8 show high jitter or loss, especially when it repeats at the same times of day.
This kind of fault even has its own place in the standards. ITU-T G.114 Amendment 2 (11/2009) added Appendix III on delay variation on unshared access lines, and it is in force. Jitter on the access line is a known, measurable fault. You're not imagining it.
What to bring:
- Wired ping results to the router and to 8.8.8.8, with the date and time.
- At least three runs at different times, including one in the evening.
- A continuous log:
ping -t 8.8.8.8on Windows (stop it with Ctrl+C) orping -c 600 8.8.8.8on a Mac, which is 10 minutes at one ping per second. - A traceroute (
tracert 8.8.8.8on Windows,traceroute 8.8.8.8on a Mac), or WinMTR or PingPlotter output showing the hop where the jitter starts.
Ask for a line check of signal levels and noise and, if the fault repeats, a technician visit. Speed test screenshots alone earn you a "your speed is fine" brush-off.
If your evidence points inside the house, go back to the fix list. If it points outside, your log is your case.
Frequently asked questions
Does jitter matter for Netflix or YouTube?
Barely. Streaming apps download several seconds of video ahead and play from that buffer, so jitter that would wreck a call goes unnoticed. Live calls and games can't buffer that far ahead, which is why they suffer.
Can jitter be zero or negative?
It can't be negative, because it measures the size of the change between packets, not its direction. Zero is rare on a real network, but a few milliseconds on a cable to your router is normal.
Does jitter lower my download speed?
Not directly. A connection can download at full speed while its jitter ruins every call, which is why a high Mbps result says nothing about call quality.