Previous work has shown that medial efferents can inhibit responses of auditory-nerve fibers to high-level sounds and that fibers with low spontaneous rates (SRs) are inhibited most. However, quantitative interpretation of these data is made difficult by effects of adaptation. To minimize systematic differences in adaptation, efferent inhibition was measured with a randomized presentation of both sound level and efferent stimulation. In anesthetized cats, efferents were stimulated with 200/s shocks and auditory-nerve-fiber responses were recorded for tone bursts (0-100 dB SPL, 5-dB steps) at their characteristic frequencies. Below 50 dB SPL, efferent inhibition (measured as equivalent attenuation) was similar for all fibers with similar CFs in the same cat. At 45-75 dB SPL, low-SR and medium-SR fibers often showed much larger inhibition, and substantial inhibition even at 100 dB SPL. Expressed as a fractional decrease in rate, at 90-100 dB SPL the inhibition was 0%, 6%, and 13% for high-, medium-, and low-SR fibers (differences statistically significant). Finding the largest equivalent attenuations at 45-75 dB SPL does not fit with the hypothesis that medial-efferent inhibition is due solely to a reduction of basilar-membrane motion. The large attenuations, some over 50 dB, indicate that medial efferent inhibition is more potent than previously reported.