When two phase-locked sinusoidal stimuli having frequencies of F1 and F2 are simultaneously introduced to the ear of the gerbil, a difference tone (DT) can be observed (DT = F2-F1, where F2 > F1) in the time-averaged electrical response recorded from the cochlear round window (RW). Tetrodotoxin (TTX), which blocks the axonal firing of the cochlear nerve fiber, greatly attenuates this DT response, suggesting it is primarily neural in origin. Alternating the polarity of a single phase-locked tone cancels out the RW cochlear microphonic (CM) from the time-averaged response, leaving a residual auditory nerve neurophonic (ANN) response if the stimulus frequency is low enough to result in phase-locked firing of cochlear nerve axons. Simultaneous presentation of 1 kHz (F1) and 2 kHz (F2) tones, each being phase-locked with alternating polarity, produces a small ANN in response to the original tones and a large time-averaged ANN in response to the DT. Even when the frequency of the individual tones is too high to support phase-locking, a large DT-ANN can also be measured in response to simultaneously presented tones. A robust time-averaged DT-ANN can be measured when the temporal and intensity relationships between F1 and F2 are varied widely, with the latency (but not amplitude) of the response following the stimulus envelope. The DT-ANN produced by pairs of tones having frequencies ranging from 500 Hz to 3.5 kHz is largest in response to a DT of approximately 700-1100 Hz. This is in contrast to the ANN generated in response to a single tone, which decreases in magnitude as the stimulus frequency increases from 500 to 1500 Hz. Robust DT-ANNs can be measured from the gerbil even when the F2 frequency is greater than 30 kHz.