Four groups of subjects comprised of one control group, two analogue phobic groups and one clinical phobic group were exposed to a dichotic listening experiment in which they were required both to shadow one message to the exclusion of the other and recognize the occurrence of target words appearing in both the attended and rejected messages. Target words were either phobic-relevant or neutral. It was found that the clinical phobic group recognized significantly more target words than any of the other groups. This difference was due to their superior recognition of phobic-relevant targets from the rejected message. The results were interpreted within a combined framework of Staats's (1968, 1975) social behaviour theory and the Rescorla and Wagner (1972) model of classical conditioning.