In statistical physics, the fluctuation dissipation theorem states that a change or fluctuation in a system will be dissipated as the system returns to equilibrium. For a small perturbation the response is linear.

Molecular scale models (microscopic models) can be used to quantitatively predict material properties in the context of linear response theory. A basic premise or theorem of statistical thermodynamics on which such predictions are based is the fluctuation dissipation theorem. It proposes that there is an explicit relationship between molecular dynamics at thermal equilibrium, and the macroscopic response that is observed in a dynamic measurement. This theorem is based on fields that are weak relative to the potential of molecular interaction so that rates of relaxation are not effected by the applied field. The essence of fluctuation-dissipation theorem is that it relates equilibrium fluctuations to out-of-equilibrium quantities, like noise power is related to resistance. "Out-of-equilibrium" in the above sentence should be understood as close to equilibrium or stationary states.