Bufferbloat test
Press Start and the test measures your ping at rest, then again while a download and an upload fill your line. Your grade comes from how much ping rises: A under 20 ms, B under 50, C under 100, D under 200, F at 200 ms or more. A or B means a busy line won't hurt games or calls. D or F means it will, and a router setting usually fixes it.
The test needs JavaScript.
- Bufferbloat grade
- –
- Ping at rest
- –
- Ping during download
- –
- Ping during upload
- –
- Download
- –
- Upload
- –
Your grade
| Grade | Ping rise under load | What it means |
|---|---|---|
| A | Under 20 ms | Ping barely rises under load. |
| B | 20 to 49 ms | A small rise; most people won't notice. |
| C | 50 to 99 ms | Noticeable lag while something else downloads or uploads. |
| D | 100 to 199 ms | Games and calls suffer whenever the line is busy. |
| F | 200 ms or more | Severe bufferbloat: any download or upload makes everything lag. |
For our verdicts, competitive gaming needs B or better and video calls need C or better. Those are our thresholds, explained on how it works. Bufferbloat grades explained has an example for each.
What bufferbloat is
Your router holds a queue (a buffer) for data waiting to go out or come in. When the line is full, as it is during a big download, packets wait in that queue. If the queue is too large and nothing manages it, a small game or call packet can sit behind seconds of download data.
That wait shows up as ping. A connection that pings 15 ms at rest can ping 300 ms while someone in the house downloads a game update, uploads photos to the cloud or starts a 4K stream. Nothing is broken and no data is lost; everything just waits. That's bufferbloat. It explains the classic complaint: "my speed test is fine, but games lag whenever my roommate is online."
A normal speed test misses it, because it reports ping before the download starts. This test measures ping while the line is busy.
How the grade is worked out
- Ping at rest. About 20 HTTP round trips to the nearest Cloudflare location, before anything else runs. The middle value (the median) is your idle ping.
- Ping during download. The test downloads from our server in growing steps. While the transfers run, it sends a ping every 400 milliseconds or so. This is called loaded latency. Your result is the median of those pings.
- Ping during upload. The same again while it uploads.
- The rise. The grade comes from the larger of the two increases over your idle ping, whichever direction is worse.
A worked example: idle ping 18 ms, 95 ms during the download and 160 ms during the upload. The upload rise is the larger one: 160 − 18 = 142 ms. That's a D. The download alone would have been a C. This pattern is common on cable, where the upload is much smaller and fills quickly.
Like every ping on this site, these are HTTP round trips, not ICMP. A browser can't send ICMP, and it can't reach game servers, so in-game ping may be higher. The rise under load is still a fair measure, because idle and loaded pings go to the same place by the same method.
The test also shows your download and upload speeds, because it has to fill the line to load it. On a fast line it can move a few hundred megabytes, so on a capped phone plan use the ping test instead.
Other bufferbloat tests use other scales
Our A to F scale is our own. Other well-known tests grade differently, so the same line can get different letters:
- Waveform's bufferbloat test runs its own download and upload and gives its own letter grade.
- LibreQoS's test measures idle and loaded latency and rates how well gaming, calls and streaming will work.
If two tests disagree by a letter, that's normal: they use different servers, different load sizes and different cutoffs. Compare before and after a change on the same test. Why results differ explains more.
How to fix a bad grade
Bufferbloat is usually fixed at the router, not by a faster plan. The fix is smart queue management (SQM), which router makers name in different ways. It keeps the queue short and shares the line fairly, so a game or call isn't stuck behind a download. Many routers have it built in. Turning it on usually means entering your line's speeds; then run this test again to check the grade.
Our guide to fixing bufferbloat goes through it step by step for common routers, with links to each maker's instructions. If your router has no such setting, routers with smart queue management lists models that do.
Some things to know first:
- Test on a cable if you can. Wi-Fi adds its own delay and makes the grade less clear (Wi-Fi vs Ethernet).
- Stop other traffic in the house. The test needs to fill the line itself; someone else's stream adds noise.
- Run it two or three times. Grades can move one letter between runs. A steady D or F is the one to act on.
- Satellite and 5G home internet often grade worse at busy times. See the guides for Starlink and T-Mobile home internet.
Sources
- Cloudflare speed test engine (@cloudflare/speedtest), README: loaded latency measured during downloads and uploads (checked November 2026)
- Bufferbloat.net: tests for bufferbloat: lists the Waveform and LibreQoS tests (checked November 2026)