During the simultaneous presentation of two primary tones to the ear (f1, f2; f2>f1), intermodulation distortion products can be measured acoustically in the ear canal and electrically as scalp-recorded auditory evoked potentials. The purpose of this investigation was to elucidate the sources of nonlinearity within the human auditory system responsible for generating the quadratic difference tone (f2- f1) and the cubic difference tone (2f1-f2). Three experiments were conducted. During the first experiment, measurements of distortion-product otoacoustic emissions and auditory evoked potentials were obtained from 24 normal-hearing adults (12 male) in conditions with and without presentation of a contralateral noise. The effects of primary-tone duration and mode of presentation of the primaries (monotic vs. dichotic) on measurements of auditory evoked potentials were examined in the second and third experiments. The results from the first experiment indicated that overall, both acoustical and electrical distortion products were suppressed during presentation of a contralateral noise. However, greater suppression occurred at the electrical cubic difference tone compared to the electrical quadratic difference tone. The second experiment revealed that growth in electrical distortion-product amplitude accompanied increases in primary-tone duration. The results of the third experiment demonstrated that electrical distortion products were more prevalent when the primaries were presented monotically. The findings from the first experiment of the investigation supported the conjecture that a cochlear nonlinearity produced acoustical and electrical cubic difference tones. Evidence from the second and third experiments concerning the origin of electrical distortion products was inconclusive, and contributions from both cochlear and neural nonlinear sources could not be ruled out.