From previous studies it appears that at least two factors limit the upper frequency at which auditory‐nerve (AN) fibers can entrain to the envelope of a sinusoidally amplitude‐modulated (AM) tone. Cochlear mechanical filtering insures that, in the local motion driving a fiber tuned to the carrier, sidetone amplitudes decrease as sidetone displacement envelopes separate with modulation frequency (fm). Only if at least one sidetone’s amplitude is large enough, relative to carrier’s, will there be modulation of basilar motion at the point tuned to the carrier. In addition, processes within haircell and fiber limit the upper frequency at which they follow variations in amplitude. To assess change, along the cochlea, in the two factors’ relative importance, AN modulation transfer functions (MTFs) [Joris and Yin, J. Acoust. Soc. Am.91, 215–232 (1992)] were replotted versus distance in mm between sidetones and carrier, using an empirical frequency‐place function [Greenwood, J. Acoust. Soc. Am.87, 2592–2605 (1990)]. MTF bandwidths, converted to mm, changed little over the apical 40% of cochlea but decreased basally. Expressed in Hz, they increased from apex to base and reached an upper limit at a characteristic frequency (CF) of 20 kHz. This is consistent with the idea that at high CFs phase‐locking constraints limit envelope‐following before mechanical filtering does, while in apical regions spatial filtering reduces envelope amplitude variation, hence envelope following, before limits on phase locking do. Consistently, MTF bandwidths parallel tuning curve bandwidths in the apical cochlea, but are smaller near the base, where tuning curve bandwidths and spatial filtering appear constant.