A key-recovery attack is an adversary's attempt to recover the cryptographic key of an encryption scheme. Normally this means that the attacker has a pair, or more than one pair, of plaintext message and the corresponding ciphertext.^[1]: 52 Historically, cryptanalysis of block ciphers has focused on key-recovery, but security against these sorts of attacks is a very weak guarantee since it may not be necessary to recover the key to obtain partial information about the message or decrypt message entirely.^[1]: 52 Modern cryptography uses more robust notions of security. Recently, indistinguishability under adaptive chosen-ciphertext attack (IND-CCA2 security) has become the "golden standard" of security.^[2]: 566 The most obvious key-recovery attack is the exhaustive key-search attack. But modern ciphers often have a key space of size ${\displaystyle 2^{128))$ or greater, making such attacks infeasible with current technology.

In cryptography, the key-recovery advantage (KR advantage) of a particular algorithm is a measure of how effective an algorithm can mount a key-recovery attack. Consequently, the maximum key-recovery advantage attainable by any algorithm with a fixed amount of computational resources is a measure of how difficult it is to recover a cipher's key. It is defined as the probability that the adversary algorithm can guess a cipher's randomly selected key, given a fixed amount of computational resources.^[3] An extremely low KR advantage is essential for an encryption scheme's security.

• cseweb.ucsd.edu paper MIT Lecture Notes on Cryptography Archived 2012-04-21 at the Wayback Machine