Guide
ML-KEM, ML-DSA and SLH-DSA: the NIST standards explained
After a public competition that ran from 2016, NIST finalized its first post-quantum standards on 13 August 2024. There are three, and they split cleanly by job: one for key establishment and two for digital signatures. Here is what each is, where it came from, and when you would reach for it.
FIPS 203: ML-KEM (key establishment)
ML-KEM, short for Module-Lattice-Based Key-Encapsulation Mechanism, is the standard for setting up a shared secret at the start of a connection. It grew out of the algorithm known during the competition as CRYSTALS-Kyber, and its security rests on the hardness of structured lattice problems.
This is the one that matters most for harvest now, decrypt later, because key establishment is the part of a session an attacker can break retroactively. It is almost always deployed in hybrid form, paired with a classical exchange such as X25519, so that a weakness in either component alone still leaves the connection protected. The hybrid group you will see negotiated on modern servers is written X25519MLKEM768.
FIPS 204: ML-DSA (the default signature)
ML-DSA, Module-Lattice-Based Digital Signature Algorithm, is the general-purpose signature standard, derived from CRYSTALS-Dilithium. Signatures prove who you are and that data was not altered. ML-DSA is the one to use for most new signing needs because it is efficient and lattice-based, matching ML-KEM.
FIPS 205: SLH-DSA (the conservative signature)
SLH-DSA, Stateless Hash-Based Digital Signature Algorithm, comes from SPHINCS+. Instead of lattices it relies only on the security of hash functions, which are very well understood. Its signatures are larger and slower, so it is not the everyday choice, but it is the conservative backup: if a weakness were ever found in lattice math, SLH-DSA would still stand.
What is still coming
NIST is deliberately keeping more than one option in each category, so a single mathematical surprise cannot undo the whole transition.
- HQC was selected in March 2025 as a second key-establishment mechanism. It is code-based rather than lattice-based, a different foundation from ML-KEM, and a draft standard is expected around 2027.
- FN-DSA (from Falcon) is a further lattice-based signature standard in progress, useful where small signatures matter.
Choosing between them
- Setting up an encrypted connection: ML-KEM, in hybrid.
- Signing, as a default: ML-DSA.
- Signing, where you want the most conservative assurance: SLH-DSA.
A readiness assessment tests for the first of these, ML-KEM key exchange, because it is both the most urgent and the part that is visible from outside a system.
See where this stands for a real domain. Type one in and get a post-quantum readiness grade from its public connections, with the reasoning behind it.
Run a scanRelated
Sources
- NIST FIPS 203 (ML-KEM)
- NIST FIPS 204 (ML-DSA)
- NIST FIPS 205 (SLH-DSA)
- NIST: HQC selected as fifth algorithm (Mar 2025)
Primary sources only. No news outlets or aggregators.