24 university students (average age 22.1 yrs) were conditioned to pictures of angry faces with a mild electric shock unconditioned stimulus/stimuli (UCS). They were then tested with backward masking conditions preventing conscious recognition of the facial stimuli. In the 1st experiment, a shock followed a particular nonmasked angry face exposed among many other faces. Although the Ss did not rate this face as familiar in a subsequent test when it was presented masked among other masked and nonmasked faces, it elicited larger skin conductance responses than did nonshocked control faces. This dissociation between explicit recognition and implicit skin conductance differentiation was replicated in the 2nd experiment, in which the Ss rated their shock expectancy. Although conditioning resulted in much better differentiation between conditioned and control faces during nonmasked than masked test trials, skin conductance differentiation did not differ between the 2 masking conditions.