The auditory evoked potential (AEP) is an electrical signal that can be recorded from electrodes attached to the scalp of a human subject when a sound is presented. The signal is considered to reflect neural activity in response to the acoustic stimulation and is a well established clinical and research tool to objectively assess the function and integrity of the auditory nervous system. However, the physiological generation of AEPs represents a complicated interaction between linear and nonlinear cochlear and neural processes and is not well understood in humans. This thesis presents and evaluates a phenomenological model of AEP generation that can predict key experimental observations of recorded AEPs. The purpose of the study was to investigate the role of the different stages of auditory signal processing and their effects on AEP generation. In recent years, there has been a push both clinically and in research towards using realistic and complex stimuli, such as speech, to electrophysiologically assess the human hearing. However, to interpret the AEP generation to complex sounds, the potential patterns in response to simple stimuli needs to be understood. Therefore, the model was used to simulate auditory brainstem responses (ABRs) evoked by classic stimuli like clicks, tone bursts and chirps. The ABRs to these simple stimuli were compared to literature data and the model was shown to predict the frequency dependence of tone-burst ABR wave-V latency and the level-dependence of ABR waveV amplitude for clicks and chirps varying sweeping rates. The model was also evaluated based on ABR recordings evoked by speech syllables, and was shown to account for the differences in the responses observed between the stimuli. It was demonstrated that the generation of the syllableevoked ABRs was highly influenced by cochlear and afferent neural processing, which supported the importance of cochlear processing for the generation of AEPs. A second major contribution of this study was the investigation of whether auditory steady-state responses (ASSRs) can be used to assess human cochlear compression. Sensorineural hearing impairments is commonly associated with a loss of outer hair-cell functionality, and a measurable consequence is the decreased amount of cochlear compression at frequencies corresponding to the damaged locations in the cochlea. In clinical diagnostics, a fast and objective measure of local cochlear compression would be of great benefit, as a more precise diagnose of the deficits underlying a potential hearing impairment in both infants and adults could be obtained. It was demonstrated in this thesis, via experimental recordings and supported by model simulations, that the growth of the ASSR amplitude with stimulus level can indeed be used as such an estimate of local cochlear compression.