These experiments add a measure of response phase variance--'phase coherence'--to the analysis procedures applied to auditory steady-state responses (SSR). The effects of stimulus frequency, intensity, rate and total number (i.e., recording time) were studied using 11 normal adult subjects. In a first experiment, SSR phase coherence was found to be highest at presentation rates near 40/sec, even when response amplitudes were higher at other rates. Further, phase coherence was observed to be linearly related (r = 0.91) to signal-to-noise ratio. Two further experiments demonstrated that phase coherence can correctly detect responses to near-threshold stimuli. In 15 min runs, significant phase coherence was detected within 6 dB of behavioral threshold in 6 subjects for 0.5 and 2.0 kHz signals, while phase coherence in no-stimulus control runs did not reach significance. Minimum data collection time required to record significant (P less than 0.01) responses was studied for 10 subjects. In 2 of 40 recordings at 10 dB SL phase coherence remained insignificant after even 15 min. However, average recording time to reach significance at 10 dB SL was less than 4 min in 38 of 40 recordings, and less than 1 min at 25 dB SL (18 of 18 recordings). These results indicate that using phase coherence to detect the presence of the 40/sec auditory steady-state response, efficient threshold search procedures may be devised to provide fast, accurate, and objective estimates of auditory behavioral thresholds in nearly all normal adults.