In each of two experiments, allocation of cognitive processing capacity was measured in college-student subjects during autonomic discrimination classical conditioning. A 7.0-sec delay paradigm was used to establish classically conditioned responses to a reinforced visual conditioned stimulus (CS+). Electrodermal responses were the primary measures of autonomic classical conditioning. Allocation of processing capacity was measured by monitoring performance on a secondary reaction-time (RT) task. The auditory secondary-task RT signal was presented before, and 300, 500, 3500, 6500, and 7500 msec following CS onset. The RT signal was also presented following properly and improperly cued shock unconditioned stimuli (UCSs). Significant discrimination classical conditioning was obtained in both experiments. Comparison with control subjects who did not receive the RT signals indicated that the presence of the RT signals did not interfere with the development of classical conditioning. Four principal findings were obtained with the secondary-task RT measure. First, RTs to signals presented during CS+ were consistently slower than RTs to signals presented during CS-. This finding indicates that greater capacity allocation occurred during CS+ than CS- and is consistent with recent cognitive interpretations of classical conditioning. Second, the largest capacity allocation (i.e., slowing of RTs) occurred 300 msec following CS+ onset. This finding is consistent with the notion that subjects are actively processing the signal properties of the CS+ at 300 msec following CS+ onset. Third, presentation of the UCS when improperly cued (following CS-) significantly increased capacity allocation, whereas presentation of the same UCS when properly cued (following CS+) did not affect capacity allocation. These findings indicate that subjects were actively prepared for the UCS following CS+ but not following CS- and that a surprising UCS elicits greater capacity allocation than does an expected UCS. Fourth, large electrodermal responders to the CSs exhibited patterns of capacity allocation during the CSs, particularly during the CS+, different from those of small electrodermal responders. In particular, they exhibited significantly longer RTs at 300 msec after CS+ onset than did the small responders, followed by a shortening of RT at 500 msec relative to the small responders. This finding suggests that large electrodermal responders devote greater processing capacity to significant environmental stimuli than do small responders and that their processing may begin and be completed more rapidly. All in all, the data indicate the complexity of the cognitive processes that occur during human classical conditioning and the usefulness of the secondary-task technique in integrating conditioning theories and psychophysiology with cognitive psychology.