In humans, the binaural interaction at the brainstem level has been studied for over 15 years. The binaural interaction component (BIC) is obtained by subtracting the summed auditory brainstem response (ABR) in the monaural stimulus mode from the ABR obtained in the binaural stimulus mode. By nature of this subtraction process, the signal-to-noise ratio of the difference waveform is poor, requiring an objective detection criterion to decide whether a significant BIC is present. In this study, the effectiveness of two analysis methods was compared. The first method is the "3 SD' method, which is based on a signal-to-noise evaluation. The second method is a template matching method, in which templates are derived from normal hearing subjects' responses and individual responses are cross-correlated with these templates. The templates were allowed to shift over a range of -0.8 to 0.8 ms in search of the maximum correlation coefficient. Thirty-nine subjects with normal hearing and five patients with a unilateral profound hearing loss participated in the study. ABRs were obtained with rarefaction and condensation clicks at a rate of 15/s and a level of 70 dB nHL. Latencies of the ABR waves I, III and V for all normal hearing subjects and for the normal ear of the patients were within the normal range. The efficiencies of both methods, defined as the number of normal hearing adults with a significant BIC plus the number of patients without a significant BIC divided by the total number of subjects, were determined. The results show that the "3 SD' method is superior to the template matching method: the efficiencies were 95% and 70% respectively, when responses to rarefaction and condensation clicks were taken together. With the "3 SD' method, a significant BIC is demonstrated in almost all normal hearing subjects (97%). However, the "3 SD' method also falsely indicated a significant BIC in one patient. These results suggest that the BIC may have clinical value in studying binaural interaction in humans.