Middle-latency responses (MLR) in humans were studied using an unconventional recording technique with wide bandpass filters. Such filtering permitted simultaneous recording of the auditory brain stem response (ABR) thus facilitating comparisons between the two responses. Effects of sedation (chloral hydrate and diazepam), stimulus-related properties and the coronal distribution of MLRs were examined. Mild sedatives did not appear to affect either MLRs or ABRs. MLRs differed from ABRs in their stimulus-related properties, implying that the neuronal mechanisms underlying their generation are not the same. The amplitude of the MLR component, Pa, was largest at the vertex and symmetrically distributed over the temporal lobes. MLR components Na and Pa and ABR wave V were reliably obtained in all subjects at moderate and high stimulus intensities. At low stimulus levels, however, the detectability of wave V was more robust than the middle-latency components. Thus ABR appears to be the test of choice when hearing sensitivity is in question. MLRs are likely to be most clinically useful in patients with neurological or central auditory processing disorders.