Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
Sweet32 is a birthday attack against 64-bit block ciphers such as 3DES and Blowfish, recovering plaintext from long-lived TLS or VPN connections that carry enough data under one key.

What Is Sweet32?
Sweet32 is a practical attack on block ciphers with a 64-bit block size, published in August 2016 by Karthikeyan Bhargavan and Gaëtan Leurent of INRIA. It does not break the cipher’s key. Instead it exploits a limit built into the block size itself: after enough blocks have been encrypted under one key in CBC mode, two of them collide, and that collision leaks information about the plaintext.
The research carries two identifiers: CVE-2016-2183 for 3DES in TLS and CVE-2016-6329 for Blowfish in OpenVPN. Both ciphers were already considered legacy in 2016, yet 3DES remained enabled on a large proportion of HTTPS servers for compatibility with Windows XP era clients.
CBC mode exclusive-ORs each plaintext block with the previous ciphertext block before encrypting. If two ciphertext blocks are ever identical, then the exclusive-OR of the two corresponding plaintext blocks equals the exclusive-OR of the two preceding ciphertext blocks, which the attacker can see on the wire. Learn one of those plaintext blocks and you learn the other.
How long until a collision? The birthday paradox says roughly the square root of the number of possible values. For a 64-bit block that is about 2 to the power 32 blocks, or 32 gigabytes of data under a single key. For a modern 128-bit cipher such as AES, the equivalent figure is 2 to the power 64 blocks, around 256 exabytes, which is unreachable in any realistic session.
That factor is the whole story. A 64-bit block size was reasonable when DES was designed in the 1970s and is simply too small for the data volumes of a modern connection.
Sweet32 is not a drive-by. The demonstrated scenario needed several things at once:
In the researchers’ proof of concept, 785 GB of traffic was captured over roughly 38 hours to recover a two-block HTTP session cookie. That is a long window, but automated clients, VPN tunnels, and database connections routinely stay open for days, and the attacker does not need to be present throughout.
Nine years on, TLS is rarely the problem. The remaining exposure tends to sit in places nobody inspects:
BF-CBC;3des-cbc or blowfish-cbc;The TLS 1.3 handshake settled the matter by removing CBC mode and all 64-bit block ciphers from the protocol entirely; only AEAD constructions such as AES-GCM and ChaCha20-Poly1305 remain. Browsers disabled 3DES suites from 2016 onwards.
NIST SP 800-131A Revision 2 restricted 3DES encryption, disallowing it for new applications and limiting the amount of data encrypted under a single key, and NIST formally deprecated it entirely. PCI DSS requires strong cryptography for cardholder data in transit, and a vulnerability scan reporting Sweet32 against an in-scope system will be treated as a failure by your assessor.
Remediation is usually straightforward because almost nothing still needs these ciphers:
testssl.sh or sslyze to find every endpoint offering 3DES or Blowfish.The NCSC’s guidance on using TLS securely recommends a short, modern cipher list rather than an inherited one, and the same discipline applies to SSH and IPsec. Our Infrastructure Security Testing Services enumerate cipher support across every listening service, including the ones that never appear in an external scan.
On public HTTPS it is largely historical, because browsers removed 3DES support and TLS 1.3 excludes 64-bit block ciphers. The risk persists on long-lived connections elsewhere: IPsec and OpenVPN tunnels, SSH sessions, and appliance management interfaces with legacy cipher lists. Those connections carry far more data under one key than a browser session ever would.
No. The attack never touches the key. It exploits the fact that a 64-bit block space is small enough for two ciphertext blocks to repeat after roughly 32 gigabytes of data, and a repeat leaks the exclusive-OR of two plaintext blocks. The attacker recovers plaintext such as a session cookie, not the key itself.
The published proof of concept captured 785 GB over about 38 hours to extract an HTTP session cookie. The theoretical threshold is around 32 GB under a single key, where collision probability becomes meaningful. Shortening rekey intervals keeps each key well below that threshold, which is a useful stopgap where a legacy cipher cannot be removed immediately.
Prefer AEAD suites with 128-bit block ciphers: AES-128-GCM, AES-256-GCM, and ChaCha20-Poly1305. Enable TLS 1.2 and TLS 1.3 only, and remove every CBC and 3DES suite from the list. For SSH, keep aes256-gcm and chacha20-poly1305 and drop the cbc variants. For IPsec, use AES-GCM proposals rather than inherited 3DES ones.
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.