Fermat's little theorem states that if p is a prime number, then for any integer a, the number a**p − a is an integer multiple of p. In the notation of modular arithmetic, this is expressed as

For example, if a = 2 and p = 7, then 27 = 128, and 128 − 2 = 126 = 7 × 18 is an integer multiple of 7.

If a is not divisible by p, Fermat's little theorem is equivalent to the statement that a**p − 1 − 1 is an integer multiple of p, or in symbols:^[1][2]

For example, if a = 2 and p = 7, then 26 = 64, and 64 − 1 = 63 = 7 × 9 is thus a multiple of 7.

Fermat's little theorem is the basis for the Fermat primality test and is one of the fundamental results of elementary number theory. The theorem is named after Pierre de Fermat, who stated it in 1640. It is called the "little theorem" to distinguish it from Fermat's last theorem.^[3]

Pierre de Fermat first stated the theorem in a letter dated October 18, 1640, to his friend and confidant Frénicle de Bessy. His formulation is equivalent to the following:^[3]

In fact, the original statement was

This may be translated, with explanations and formulas added in brackets for easier understanding, as:

Fermat did not consider the case where a is a multiple of p nor prove his assertion, only stating:^[4]

Euler provided the first published proof in 1736, in a paper titled "Theorematum Quorundam ad Numeros Primos Spectantium Demonstratio" in the Proceedings of the St. Petersburg Academy,^[6] but Leibniz had given virtually the same proof in an unpublished manuscript from sometime before 1683.^[3]

The term "Fermat's little theorem" was probably first used in print in 1913 in Zahlentheorie by Kurt Hensel:

An early use in English occurs in A.A. Albert's Modern Higher Algebra (1937), which refers to "the so-called 'little' Fermat theorem" on page 206.^[7]

Some mathematicians independently made the related hypothesis (sometimes incorrectly called the Chinese Hypothesis) that 2p ≡ 2 (mod p) if and only if p is prime. Indeed, the "if" part is true, and it is a special case of Fermat's little theorem. However, the "only if" part is false: For example, 2341 ≡ 2 (mod 341), but 341 = 11 × 31 is a pseudoprime. See below.

Several proofs of Fermat's little theorem are known. It is frequently proved as a corollary of Euler's theorem.

A corollary of Euler's theorem is: for every positive integer n, if the integer a is coprime with n then

If n is prime, this is also a corollary of Fermat's little theorem. This is widely used in modular arithmetic, because this allows reducing modular exponentiation with large exponents to exponents smaller than n.

If n is not prime, this is used in public-key cryptography, typically in the RSA cryptosystem in the following way:^[8] if

Fermat's little theorem is also related to the Carmichael function and Carmichael's theorem, as well as to Lagrange's theorem in group theory.

The converse of Fermat's little theorem is not generally true, as it fails for Carmichael numbers. However, a slightly stronger form of the theorem is true, and it is known as Lehmer's theorem. The theorem is as follows:

If there exists an integer a such that

and for all primes q dividing p − 1 one has

then p is prime.

This theorem forms the basis for the Lucas–Lehmer test, an important primality test.

If a and p are coprime numbers such that a**p−1 − 1 is divisible by p, then p need not be prime. If it is not, then p is called a (Fermat) pseudoprime to base a. The first pseudoprime to base 2 was found in 1820 by Pierre Frédéric Sarrus: 341 = 11 × 31.^[9][10]

A number p that is a Fermat pseudoprime to base a for every number a coprime to p is called a Carmichael number (e.g. 561). Alternately, any number p satisfying the equality

is either a prime or a Carmichael number.

The Miller–Rabin primality test uses the following extension of Fermat's little theorem:^[11]

This result may be deduced from Fermat's little theorem by the fact that, if p is an odd prime, then the integers modulo p form a finite field, in which 1 has exactly two square roots, 1 and −1.

The Miller–Rabin test uses this property in the following way: given p = 2s d + 1, with d odd, an odd integer for which primality has to be tested, choose randomly a such that 1 < a < p; then compute b = a**d mod p; if b is not 1 nor −1, then square it repeatedly modulo p until you get 1, −1, or have squared s times. If b ≠ 1 and −1 has not been obtained, then p is not prime. Otherwise, p may be prime or not. If p is not prime, the probability that this is proved by the test is higher than 1/4. Therefore, after k non-conclusive random tests, the probability that p is not prime is lower than (3/4)k, and may thus be made as low as desired, by increasing k.

In summary, the test either proves that a number is not prime, or asserts that it is prime with a probability of error that may be chosen as low as desired. The test is very simple to implement and computationally more efficient than all known deterministic tests. Therefore, it is generally used before starting a proof of primality.