This paper presents a classifier system based on a stochastic associative memory using single-electron basic circuits. This memory can stochastically extract, from a set of stored reference patterns, the one that is most similar pattern to a given input pattern. Simulations are carried out to demonstrate the functionality of the system.