The use of a stochastic resonance model contributes crucially to our comprehension of the intensity resolution characteristics of the mammalian cochlea. In guinea pigs, as demonstrated by different statistical methods, the temporal distribution of the interspike intervals of the spontaneous activity reflects an intrinsic cochlear white noise process, demanded as basic requirement for manifest stochastic resonance phenomena. Brownian motion of cochlear fluids is discussed as the underlying white noise motor. Following our model, the amount of white noise, adjusted at the level of the stereocilia of the inner hair cells, determines the threshold, dynamic range and intensity discrimination limen of an individual afferent neuron of the mammalian cochlea.