The auditory brain stem response (ABR) and middle latency response (MLR) were studied in 48 young children (96 ears). The responses were elicited using low intensity stimuli (30-dB nHL clicks) and simultaneously were recorded on a dual time base. Both the ABR and MLR were elicited in 70 ears. In 12 ears, just one response was recorded (ABR in eight ears and the MLR in four ears). In 14 ears, neither response was recorded. Test-retest analysis on the same subject demonstrated that the ABR was more repeatable and easier to identify than the MLR. The test-retest difference was determined for the amplitude and latency of the ABR and MLR waveforms. The test-retest latency difference for wave Pa was found to be 3.6 times larger than for wave V. The normalized test-retest amplitude difference for P phi-Na, Na-Pa, and Pa-Nb was found to be two to three times larger than for wave V. These data support the conclusion that the ABR, rather than the MLR, should be used to measure hearing in young children. The authors also advocate using minimal high pass (HP) filtering when recording the ABR in a sedated or sleeping child. Muscle artifact was not found to be a problem. The authors suggest the use of minimal HP filtering so that phase-shift distortion is minimized and a larger response amplitude can be recorded.