The present experiment examined the effects of respondently conditioned emotional functions on the formation of stimulus equivalence relations. Fifty-seven participants were exposed to a stimulus-pairing procedure that paired six nonsense syllables with aversive images, and a further six stimuli with neutral images. A second phase established different operant response functions for one aversive CS and one neutral CS. In Phase 3, 45 of the 57 participants demonstrated a transfer of the established operant response to stimuli sharing respondent functions, thereby demonstrating the formation of two functional classes. Using a between-subjects design, participants were then exposed to a conditional discrimination training and testing protocol designed to establish two three-member stimulus equivalence relations using either six aversive or six emotionally neutral CSs as stimuli. Participants required significantly more testing trials to form stimulus equivalence relations when all stimuli had emotionally aversive functions compared to neutral functions. Implications of this study for the treatment of clinical anxiety are considered.