Brain-stem auditory evoked potentials (BAEPs) to paired stimuli of various intervals were recorded using a subtraction technique to cancel the first and facilitate recognition of the second response. In a pilot series of 12 experiments, no latency change was encountered with paired click intervals (PCI) of 8 and 6 ms. Thirty healthy subjects were then investigated by presenting single and paired stimuli with a PCI of 4, 3.1, 2.3 and 1.5 ms at a repetition rate of 10/s to each subject. With diminishing PCIs, identification and latency measurement of waves IV and V became increasingly difficult because of marked amplitude reduction. Component III remained identifiable in all but two recordings and showed a statistically significant latency increase and amplitude reduction only when the stimuli were presented 1.5 ms apart. Component I did not show a consistent latency change with any PCI but its amplitude was significantly reduced with 4 ms PCI. These findings differ from those reported when using PCIs of 5 ms or more and when using single stimuli at a high repetition rate, where component V was the most stable. The alterations of evoked potentials found only with a PCI of 1.5 ms are probably of central origin.