There is now considerable evidence that phobic responding is associated with a bias towards expecting aversive or traumatic outcomes following encounters with the phobic stimulus (e.g. Davey, 1995). In terms of conditioning contingencies, this can be described as a bias towards expecting an aversive or traumatic outcome (the unconditioned stimulus - UCS) following a phobic stimulus (the conditioned stimulus - CS). Research into the role of UCS expectancy biases in the development and maintenance of specific phobias has three basic requirements. First, it is not clear whether the ues expectancy biases evidenced in specific phobias represent a stimulus specific response or a more generalised associative phenomenon. Second, it is not clear what dispositional or state factors might contribute to the development and maintenance of such ues expectancy biases. Third, it is not clear what type of cognitive mechanisms might underlie UCS expectancy biases. This thesis uses a thought experiment version of a threat conditioning procedure to explore these requirements. The key findings indicate that spider phobics tend to overestimate the likelihood of aversive outcomes following phobic, but not other fear relevant stimuli, and tend to underestimate the likelihood of aversive outcomes following fear irrelevant stimuli in comparison to non-phobic controls. This dichotomous ues expectancy bias is mirrored both in the evaluation of stimuli in terms of dangerousness and valance, and in the generation of harm and safety cues with regards to these stimuli. Both positive and negative mood states, but not arousal states contributed to ues expectancy inflation, and in the case of revulsive animals induced state disgust also increased reported ues expectancies. The thesis concludes with an evaluation of the role of UCS'S expectancies in the development and maintenance of specific phobias, and a discussion of the implications of these findings for our understanding of the information processing mechanisms underlying the specific phobias.