The one-time pad is a cipher in which each message is combined with a key that is truly random, at least as long as the message itself, and never reused. Under these conditions it is the only encryption scheme proven to provide perfect secrecy: the ciphertext reveals nothing whatsoever about the plaintext to anyone who does not hold the key.
That proof comes from Claude Shannon’s 1949 paper “Communication Theory of Secrecy Systems,” published in the Bell System Technical Journal. Shannon defined perfect secrecy as the condition that the probabilities of the possible messages, given the ciphertext, are unchanged from their probabilities before the ciphertext was seen. He showed this property is realized by the Vernam system, the one-time pad, when the key is random and as long as the message.
The strength of the scheme is also its limitation. Perfect secrecy requires that the key contain at least as much randomness as the message, so two parties who want to exchange a megabyte of secret traffic must first share a megabyte of secret key. Distributing and safely storing that much key material, and never reusing any of it, is the practical burden that keeps the one-time pad out of everyday use.
Reuse is fatal: the moment a pad is used for a second message, the perfect-secrecy guarantee collapses and an attacker can begin recovering plaintext. For this reason real systems instead rely on shorter keys with computational rather than information-theoretic security, but the one-time pad remains the theoretical benchmark against which all other ciphers are measured.