Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
POODLE exploits the way SSL 3.0 handles CBC padding, letting an attacker who can force a protocol downgrade recover a session cookie one byte at a time.

What Is the POODLE Attack?
POODLE—Padding Oracle On Downgraded Legacy Encryption—was published by a Google research team on 14 October 2014 as CVE-2014-3566. It showed that SSL 3.0, a protocol already fifteen years old and superseded three times over, could be used to decrypt the secret parts of an encrypted session byte by byte. The attack required the attacker to be on the network path and to run script in the victim’s browser, but both conditions are met routinely on a public wireless network.
When a block cipher such as 3DES or AES in CBC mode encrypts data, the plaintext must be padded to a multiple of the block size. TLS specifies exactly what those padding bytes must contain and verifies them on decryption. SSL 3.0 does not: it specifies only the final byte, which gives the padding length, and leaves the preceding padding bytes undefined. They are not covered by the message authentication code either.
That gap creates an oracle. An attacker who modifies a ciphertext block and sends it to the server learns something from the response: if the server accepts the record, the forged padding happened to be valid, which reveals a byte of plaintext. On average 256 attempts recover one byte.
The practical version works like this. The attacker uses JavaScript on a page the victim is visiting to make repeated requests to the target site, each time adjusting the request path length so that a chosen byte of the session cookie lands in the last position of a block. They then swap in a block from elsewhere in the stream and observe whether the server accepts it. With enough requests—the paper estimated around 256 per byte—the entire cookie falls out, and the attacker uses it for session hijacking.
By 2014 almost no client preferred SSL 3.0. The attack worked because of downgrade dances: when a TLS handshake failed, browsers would silently retry with a lower protocol version to accommodate badly behaved middleboxes and servers. An attacker in the path could simply cause the first handshakes to fail, and the browser would obligingly fall back to SSL 3.0 on its own.
This behaviour was the deeper problem, and it is why the mitigation came in two parts: remove SSL 3.0 support, and implement TLS_FALLBACK_SCSV (RFC 7507), a signalling mechanism that lets a client tell the server “this is a fallback attempt”, so a server supporting better protocols can refuse it.
Two months later, in December 2014, a related issue appeared as CVE-2014-8730. Several TLS implementations—notably in F5 and A10 load balancers—had copied SSL 3.0’s lax approach and failed to check the padding bytes even while speaking TLS 1.0 or later. These were vulnerable without any downgrade at all, which made them worse in practice. The fix was a vendor patch rather than a configuration change.
The remediation has been settled for a decade and is purely configuration:
PCI DSS required organisations handling cardholder data to migrate off SSL and early TLS by 30 June 2018, so this is also a compliance item for any UK business in scope of PCI DSS. Modern browsers removed SSL 3.0 support entirely during 2015, which means the remaining risk sits in server-to-server and appliance traffic rather than in user browsing.
POODLE is a padding oracle, the same family as Vaudenay’s 2002 work and the Lucky Thirteen timing attack of 2013. The recurring pattern is that encrypt-then-MAC is safe and MAC-then-encrypt invites trouble, which is why TLS 1.3 removed CBC cipher suites altogether. The operational lesson is simpler still: a protocol version you keep “just in case something old needs it” is a protocol version an attacker can select for your users. Routine configuration review as part of vulnerability scanning catches this cheaply, and our Network Security Services cover protocol configuration across the whole external estate rather than web endpoints alone.
Rarely on public web servers, but it still appears on older mail servers, VPN concentrators, printers, and device management interfaces that were configured years ago and never reviewed. Internal services are the most common place scanners find it, because they are excluded from external testing. Any scan that covers only ports 443 and 8443 will miss most of the remaining instances.
Whatever sits at a predictable position in the encrypted stream, which in practice means the session cookie or authentication header. The attack does not recover the private key or decrypt the whole session at once; it extracts one byte per few hundred requests. That is sufficient, because a stolen session cookie grants the attacker the victim’s authenticated session directly.
The original attack targets SSL 3.0 specifically. A related issue, CVE-2014-8730, affected TLS implementations that failed to validate padding bytes correctly, including some F5 and A10 products. Those required vendor patches rather than configuration changes, and they were more serious because no protocol downgrade was needed for exploitation.
Scan each service with testssl.sh, sslyze, or nmap’s ssl-enum-ciphers script, which will report the supported protocol versions. For public web endpoints the Qualys SSL Labs test gives a clear grade. Include mail servers, VPN gateways, databases, and management consoles in the scope, since those are where legacy protocol support tends to survive unnoticed.
You May Be Interested In These
We use a few cookies that are needed to make this website work, and they are always on. We would also like to set analytics cookies, to see which pages are useful; functional cookies, so that our live chat works; and advertising cookies, to measure our campaigns. We will not set any of those until you say yes, and you can change your mind at any time from the Cookie Settings link in the footer.
Choose the cookies you are happy for us to use. Strictly necessary cookies are always on; everything else is off until you turn it on.
These cookies keep the website working: they hold your security checks, your form submissions and the choice you make here. The site cannot run without them, so they are always on.
These let our live chat open and remember your conversation, and let the newsletter sign-up work. Switch them off and the chat window will not appear.
These tell us which pages are read and how people find us, so that we can improve the site. The reports are about groups of visitors, never about you personally.
These let us measure our advertising and show you our adverts on other websites, such as LinkedIn, Google and Meta. Switch them off and you will still see adverts, but they will be less relevant.