The absolute sensitivity of 3‐month‐old human infants was estimated using ABR and a behavioral measure. The stimuli were tone pips with 2‐cycle rise, no steady state, and 2‐cycle fall. Frequencies of 1, 4, and 8 kHz were used. For the ABR, thresholds were determined by reducing the level of the tone pip from 20 dB nHL until the response was no longer observed, and then increasing the level until a response was again seen. Threshold was defined as the lowest level at which the ABR could be observed. Behavioral thresholds were determined using an adaptive procedure. Threshold was defined as the average of the last eight reversals obtained. A trial was considered “correct” if an observer said “yes” when a signal was presented to the infant, or ”no” when no signal was presented. The observer had no prior knowledge of the type of trial being presented, and none of the adults in contact with the infant could hear the tone pips. The correlation between ABR and behavioral thresholds depended on the frequency of the tone pip. This suggests that different factors contribute to between‐subject variability in the two measures, but that under certain circumstances, infant behavioral threshold can be predicted from the ABR threshold with some degree of accuracy. [Work supported by NIH DC000520.]