Abstract: Used pictures of potentially phobic objects as CSs for electrodermal responses to examine the validity of the equipotentiality premise in human classical conditioning. Three experiments with a total of 174 undergraduates involved a long interstimulus-interval differential conditioning paradigm with different pictures as CSs and electric shock as the UCS. In Exp I different pictures proved to be differentially effective as CSs. A group conditioned to potentially phobic stimuli (snakes or spiders) showed greater resistance to extinction than a group conditioned to fear-irrelevant pictorial stimuli (flowers or mushrooms). A 3rd group conditioned to "representative laboratory stimuli" (circles or triangles) fell in between these groups. Exp II showed differences in salience did not produce similar effects to those observed with phobic and fear-irrelevant stimuli in Exp I. In Exp III superior resistance to extinction for phobic stimuli was demonstrated when the UCS was an electric shock, but not when it was a tone to which the S produced reaction times. Thus, the effect appears specific for aversive UCSs, and CS-UCS "belongingness" has been demonstrated. It is concluded that findings challenge the premise of equipotentiality in human conditioning; data seem best explained in terms of biologically oriented constructs, such as preparedness.