Distortion‐product otoacoustic emission (DPE) input/output (I/O) curves were measured at 2f1−f2frequencies with spectral‐averaging and time‐averaging procedures from ten normal‐hearing ears. Stimuli were equal‐levelf1andf2primary tones withf2/f1ratios of 1.2 andf2frequencies at 1200, 2400, and 4800 Hz. Time‐averaging procedures lowered noise‐floor (NF) levels, compared to spectral averaging, so DPEs could be elicited by stimuli as low as 18 dB SPL in some ears. DPEs were corrected for power summation with NF levels and the resulting DPE I/O curves were fit with sixth‐order polynomials. Slopes of fitted I/O curves were specified by the first derivative as a function of the level of the primary tones. Slopes of uncorrected spectrally averaged DPE I/O curves were strongly influenced by the NF level. Slopes of NF‐corrected spectrally averaged DPE I/O curves were more representative of the true slope, specified as the slope of the corresponding time‐averaged DPE I/O curve well above its NF. True DPE I/O slopes decreased with level from a mean slope near 1.0 dB/dB at 30 dB SPL and below for all threef2frequencies. Nonmonotonicities in I/O curves werenotseen at very low stimulus levels. At moderate stimulus levels and above, nonmonotonicities in individual DPE I/O curves reduced the slopes of individual I/O curves and reduced the average slope across ears.