Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
TLS 1.3 is defined as much by what it deletes as by what it adds: static RSA, CBC ciphers, compression, renegotiation, and every legacy option that produced a decade of attacks.

What Was Removed in TLS 1.3?
RFC 8446 took four years and twenty-eight drafts, and the most consequential work in it was subtraction. Every major TLS vulnerability between 2009 and 2017—BEAST, CRIME, Lucky Thirteen, POODLE, FREAK, Logjam, DROWN, ROBOT—exploited an option that existed for backwards compatibility rather than a flaw in the core design. TLS 1.3, published in August 2018, removed the options.
The result is a protocol with a cipher suite list of five entries instead of hundreds, and far fewer ways to configure it badly.
In TLS 1.2 a client could encrypt the premaster secret under the server’s RSA public key. That meant anyone holding the server’s private key—now or in ten years—could decrypt any recorded session. TLS 1.3 removes it entirely. Every handshake uses ephemeral Diffie-Hellman over a named group, so forward secrecy is mandatory rather than optional.
Arbitrary Diffie-Hellman parameters went with it. A server can no longer offer a custom group of its own choosing, which is what made Logjam possible in 2015: clients could be downgraded to 512-bit export-grade groups whose parameters were precomputable. TLS 1.3 permits only a fixed list of vetted curves and finite-field groups.
| Removed | Attack it enabled |
|---|---|
| CBC-mode cipher suites | BEAST (2011), Lucky Thirteen (2013), POODLE (2014), padding oracles generally |
| RC4 | Biased keystream, practical plaintext recovery from 2013 onwards |
| 3DES | Sweet32 birthday attack on the 64-bit block size (2016) |
| Export and NULL cipher suites | FREAK (2015) |
| MD5 and SHA-1 signature algorithms | SLOTH transcript collision attacks (2016) |
| Static RSA key exchange | DROWN (2016), ROBOT (2017) |
What remains is authenticated encryption with associated data: AES-GCM, AES-CCM, and ChaCha20-Poly1305. The cipher suite identifier itself was redefined, listing only the AEAD algorithm and the hash, because key exchange and signature choices moved into separate extensions.
With the legacy paths removed the handshake fits into one round trip rather than two, which is a visible latency win on mobile networks. The server certificate is now encrypted, so a passive observer sees only the Server Name Indication—and Encrypted Client Hello is closing that gap too. Version negotiation moved out of the legacy_version field into a supported_versions extension, because middleboxes had made the original field unusable. A sentinel value in the ServerHello random detects downgrade attempts. The key schedule was rebuilt around HKDF with distinct secrets for each stage.
The one addition that adds risk rather than removing it is 0-RTT resumption, which lets a client send application data in its first flight. Those early bytes are replayable, so they must carry only idempotent requests. Many deployments disable 0-RTT for anything that changes state.
Organisations that decrypted internal TLS passively, by giving an inspection appliance a copy of the server’s RSA private key, lost that ability—there is no static key to share. The proposed Enterprise Transport Security variant, which reintroduced static keys, was rejected by the IETF and by browser vendors. The supported approach is now an inline proxy that terminates and re-establishes the connection, or endpoint-based visibility. Legacy middleboxes that parsed handshakes they did not understand also caused real breakage during rollout, which is why GREASE values are now injected deliberately to keep implementations tolerant of unknown parameters.
The NCSC recommends TLS 1.3 for new services, and PCI DSS has required the removal of early TLS for years. If your estate still negotiates CBC suites or static RSA, the TLS configuration review in our Infrastructure Security Testing Services will find them.
Yes, and most organisations should for now. TLS 1.2 is secure when restricted to AEAD cipher suites with ephemeral key exchange, and some clients and partner integrations still require it. Support both, disable everything older, and remove the weak TLS 1.2 options rather than the version itself. Plan the final retirement once your traffic analysis shows 1.3 everywhere.
Purely for compatibility. Middleboxes deployed in the 2010s inspected handshakes and dropped anything that did not match the TLS 1.2 shape, which broke early TLS 1.3 deployments at significant rates. The specification therefore retains a dummy ChangeCipherSpec record that has no cryptographic meaning but makes the exchange look familiar enough to pass through.
It prevents passive decryption using a copy of the server’s private key, because no such key is used for key exchange. Inspection is still possible with an inline proxy that terminates the client connection and opens its own to the server, or by collecting telemetry on the endpoints themselves. Both approaches have privacy and performance implications worth documenting before deployment.
Only for requests that can be repeated without effect. Early data is not forward secret against replay, so an attacker who captures it can send it again. Restrict 0-RTT to idempotent operations such as GET requests for public content, reject it for authenticated state changes, and apply a single-use ticket or anti-replay window at the server.
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.