English 
French 
When Cloudflare has a bad hour, it can look like half the internet broke at once. Sites you use every day throw error pages, and some of those pages mention Cloudflare by name. That’s the moment people search “is Cloudflare down.”
The useful question is narrower: is Cloudflare failing, or is one website’s setup failing? The two look alike from the outside and need different responses.
Check if Cloudflare is down right now
Run the site that’s giving you trouble through CheckSiteStatus. Test the failing site, not cloudflare.com. Cloudflare’s own marketing site can load fine while its network struggles, so a green result for cloudflare.com tells you very little.
Then compare against Cloudflare’s official status page at cloudflarestatus.com. If the status page lists an active incident and your test fails, Cloudflare is the likely cause. If the status page is clean and the test still fails, the problem is more likely that one site’s server or configuration. Our guide to how status pages work and where they lag explains why a clean page in the first few minutes isn’t proof of anything.
Read the error page, it tells you where the break is
Many Cloudflare error pages show three boxes in a row: your browser, Cloudflare, and the website’s host server. Each box carries a status. Whichever one is marked as an error is where the failure sits.
- Cloudflare box marked as an error: the problem is on Cloudflare’s side.
- Host box marked as an error: Cloudflare is working and the site’s own server is failing.
- Browser box marked as an error: the problem is on your end.
The page also shows a Ray ID, a long string at the bottom. Site owners can give it to Cloudflare support to trace the exact request. If the error code is in the 520 to 526 range, our breakdown of 521, 522 and 523 Cloudflare errors covers what each one means.
What recent Cloudflare outages looked like
Cloudflare publishes post-mortems for its larger incidents, which makes the pattern easy to see.
November 18, 2025. A bug in the logic that generated a Bot Management feature file hit many services at once. Early statements pointed to unusual traffic, but the post-mortem named the file bug. Sites including ChatGPT and X went offline for a while.
December 5, 2025. A roughly 25-minute outage starting around 08:47 UTC. Cloudflare had changed its Web Application Firewall to mitigate a vulnerability in React Server Components, and visitors saw 500 errors. The company said it was not an attack.
January 2026. On January 14, a 1.1.1.1 change reordered CNAME records in DNS responses and broke resolution for some clients. On January 22, an automated routing policy error leaked BGP prefixes from a Miami data center.
February 20, 2026. The longest of the year so far. A change to how Cloudflare manages customer IP addresses caused about 1,100 Bring Your Own IP prefixes to be withdrawn from the internet. The incident ran 6 hours and 7 minutes, mostly spent restoring configurations. Cloudflare said it was not caused by a cyberattack.
August 2026. The status page logged 13 incidents between August 7 and 14, nearly all rated minor, touching storage, workers, email security and regional network performance.
Notice what these share. The big ones trace back to Cloudflare’s own configuration changes, not to attackers. In May 2026 the company announced it had finished a program called Code Orange: Fail Small, meant to make risky changes roll out in smaller, safer steps. The August streak shows minor incidents didn’t stop, so treat that as a goal rather than a guarantee.
Why one provider’s bad day hits so many sites
Cloudflare sits in front of a very large share of the web. A dataset of 21,325 large organizations collected in August 2026 found it in front of 59.2% of them, more than any other CDN or cloud provider. When the layer in front of your site fails, the site looks dead even though your server never received the request. Our hosting provider status hub lists the other layers worth checking when a site goes down.
If you’re just a visitor
There’s nothing to fix. Clearing your cache or restarting your router won’t help if Cloudflare is the problem. Wait a few minutes and retry. If other sites that use Cloudflare also fail, you’ve confirmed it. For a general walkthrough, see our guide on telling if a site is down for everyone or just you.
If it’s your site
- Check cloudflarestatus.com and subscribe to updates for the components you use.
- Read the error code. A 52x error often means Cloudflare is fine and your origin server is the issue.
- Test your origin directly, bypassing Cloudflare, to see if the server itself responds.
- As a last resort, set the DNS record to DNS only so traffic skips Cloudflare. This removes its protection and can expose your server’s IP address, so use it only if your origin is healthy and you accept the risk.
- Set up monitoring that runs from outside Cloudflare, so you hear about an outage from an alert and not a customer. Our uptime alerts setup guide walks through it.
Test the affected site with CheckSiteStatus and compare it with cloudflarestatus.com. A failing test plus an active incident points to Cloudflare.
The site uses Cloudflare, and either Cloudflare or that site’s server is failing. The three-box diagram on the error page shows which.
Â
Most are short. December 5, 2025 lasted about 25 minutes. The February 20, 2026 incident ran just over 6 hours.
If several unrelated sites fail at once, it’s Cloudflare. If only one fails and the rest load, that site’s setup is the likelier cause.
No. Cloudflare said the December 2025 and February 2026 outages were not attacks, and the larger ones this year came from its own configuration changes.