Spontaneous, simultaneous evoked as well as delayed evoked emissions are studied in a hardware model of peripheral preprocessing with nonlinear feedback. The results suggest many very close parallels to the data found for otoacoustic emissions in man. From these parallels and the additional data measurable only in the model, it can be concluded that: (a) the cochlea acts in a similar way as established in the model; (b) the three kinds of emissions stem from the same source; (c) the phase response of the cochlea’s hydromechanics is responsible for the frequency distance between neighboring emissions as well as for the additional tips in suppression tuning curves; (d) the long delay of the delayed evoked emissions is due to the many decaying contributions from the places along the basilar membrane which cancel each other in the early part but sum up to the delayed emission in the later part; and (e) the double‐peaked shape of the suppression‐period patterns produced by high‐level, low‐frequency tones reflects the symmetrically shaped saturating nonlinearity of the feedback loops in the model which correspond to the function of the outer hair cells.