Human listeners are able to concentrate on listening to one voice amidst other conversations and background noise, but not all of the neural mechanisms for this process are understood. There is growing evidence in normal-hearing subjects that the medial olivocochlear (MOC) auditory efferent system is involved in the detection of signals in noise, such as speech sounds, by modulation of cochlear active physiological mechanisms. The present investigation aimed to evaluate the MOC efferent involvement in speech intelligibility in noise and spatial release from masking (RFM) in normalhearing adults of different ages. Contralateral suppression (CS) of distortion product otoacoustic emission was used to measure MOC efferent system function. Using HINT (Hearing in the Noise Test), we measured speech intelligibility in noise at 0 degree azimuth (HINT N0) and the improvement of speech intelligibility in noise, i.e. release from masking (RFM), when speech and noise were spatially separated. Correlation analysis was applied to reveal relations between the MOC efferent system, speech intelligibility in noise and spatial RFM. The findings suggest: (1) age-related difficulty understanding speech in background noise is related to an age-related functional decline of the MOC efferent system, (2) the higher frequency (4-6 kHz) range of the MOC efferent function is correlated with speech processing in background noise, and (3) the 1- 2 kHz frequency range of the MOC efferent system is correlated with a spatial RFM, i.e., "cocktail party" processing capability based on binaural hearing. In conclusion, the MOC efferent system can be characterized as a nonlinear adaptive filter activated during speech processing in background noise and also as a cocktail party processor.