Bats use natural landmarks such as trees for orientation. Echoes reflected by a tree are stochastic and complex. The degree of irregular loudness fluctuations of perceived echoes, i.e. the echo roughness, may be used to classify natural objects reliably. Bats are able to discriminate and classify echoes of different roughness. A neural correlate of the psychophysical roughness sensitivity has been described in the auditory cortex of the bat Phyllostomus discolor. Here, the role of the inferior colliculus of P. discolor is explored in the neural representation of echo roughness. Using extracellular recording techniques, responses were obtained to simulated stochastic echoes of different roughness. The representation of these irregular loudness fluctuations in echoes is compared to the representation of periodic loudness fluctuations elicited by sinusoidal amplitude modulation (SAM) and to the shape of the peri-stimulus time histogram in response to pure tones. About half the recorded units responded significantly differently to echoes with different roughness. Roughness sensitivity was related to the units' sensitivity to the depth of an SAM: units that responded best to strong SAMs also responded best to echoes of high roughness. In response to pure tones, these units were typically characterized as Onset units. In contrast to the auditory cortex experiments, the responses of many units in the inferior colliculus decreased with increasing echo roughness. These units typically preferred weak SAMs and showed a sustained response to pure tones. The data show that auditory midbrain sensitivity to SAM is an important prerequisite for the neural representation of echo roughness as an ecologically important echo-acoustic parameter.