These experiments were designed to test the hypothesis that glides in frequency are detected and discriminated by monitoring changes in excitation level on the low-frequency side of the excitation pattern. Thresholds were measured for detecting an increase in the extent of a frequency glide, for various standard extents (transition spans). The center frequency of each stimulus was roved, to prevent subjects from using the start or endpoint frequencies of the stimuli as cues. The level was either fixed at 70 dB SPL, or changed linearly in dB/s by an amount that varied randomly in extent and direction, keeping the level at the midpoint of the glide at 70 dB SPL. These random changes in level were intended to disrupt cues based on monitoring changes in excitation level on one side of the excitation pattern. For some conditions, performance was too good to be explained by subjects monitoring the start or endpoint frequencies of the stimuli. Performance was also too good to be explained in terms of the discrimination of changes in excitation level on one side of the excitation pattern. Thresholds, expressed as a proportion of the equivalent rectangular bandwidth (ERB) of the auditory filter, did not vary greatly with center frequency (0.5, 2, or 6 kHz), suggesting that discrimination did not depend strongly on information derived from phase locking. Glide duration (50 or 400 ms) and glide direction (upward or downward) also had little effect. Thresholds increased with increasing standard transition span, when that span was increased beyond 0.5 ERB. It is concluded that changes in glide extentper secan be discriminated, but this is not done by monitoring just one side of the excitation pattern.