An important characteristic of human hearing is that it amplifies weak sounds while attenuating louder ones. This gain transformation takes place in the inner ear (i.e., cochlea), and is responsible for a compressive relation between the level of the presented and perceived sound. The cochlear gain mechanism is essential for our hearing and degrades when hearing impairment develops. A comprehensive understanding of the gain involved in the intact human cochlea is crucial, as hearing instruments try to compensate for the loss in cochlear gain caused by hearing damage. This thesis investigates dynamic, or time-dependent, properties of cochlear gain. A time scale from 0 to 10 ms is considered to ensure that cochlear processing is investigated without including influences from higher stages in the brain. The results are expected to provide insight into how e.g. onsets of sounds are processed by the intact human system. Click-evoked otoacoustic emissions (CEOAEs) were used as a non-invasive technique to investigate cochlear gain mechanisms. CEOAEs are echoes to click stimuli that can be recorded in the ear canal, and that are produced in the cochlea as a byproduct of the nonlinear gain mechanism. Experimental results demonstrated that the CEOAE level-curve, i.e. the relation between click and CEOAE level, altered when a click was presented close in time to the evoking click. This effect was named "temporal adaptation" of the CEOAE level-curve, and was shown to operate on a time scale of 0 to 8 ms. The relation between dynamic features of CEOAEs and the underlying cochlear gain mechanism was furthermore investigated by means of a numerical model of the cochlea that simulates CEOAEs. The simulations provided insight into level-dependent features of the cochlear gain mechanism that underlie the generation of the CEOAE. In order to account for key features of temporal adaptation in CEOAEs, the existence of a time dependence in the cochlear gain mechanism was suggested. Overall, this study has demonstrated that cochlear compression characteristics can change on a time scale of 0–8 ms. The existence of such a time constant in cochlear compression may be of interest for the future development of signal processing in hearing instruments.