This study addressed the hypotheses that the growth of the level of di
stortion product otoacoustic emissions (L(DP)) With primary-tone level
reflects the behavior of a third-order nonlinear polynomial system, a
nd that two sources exist for these distortion products. The results i
ndicated that the 2f(1)-f(2) otoacoustic emission in humans can be mea
sured over a much larger stimulus range than reported previously, even
for stimuli (L(1) = L(2)) as low as 10 dB SPL (re 20 mu Pa). The inpu
t/output functions are best described as a straight line with a rate o
f growth of about 1 dB/dB of stimulus level. For stimulus levels at wh
ich metabolically active, nonlinear cochlear processes are in operatio
n, the system does not behave as a simple third-order nonlinear polyno
mial. Small plateaus and sharp discontinuities or 'notches' can occur
in the functions at stimulus levels of approximately 55 dB SPL. These
characteristics are consistent with the notion of two separate sources
of the L(DP), one at low stimulus levels, and one at high levels. An
alternative explanation is that the measured otoacoustic emission does
not represent only the activity ata single location along the basilar
membrane but includes the effects of interactions among similar signa
ls arising from multiple locations, or from the original source via mu
ltiple paths.