Twenty-four anxiety disorder subjects. 12 simple phobics and 12 panic disorders with agoraphobia, were assessed for physiological response (SCL, HR and EMG) to phobic imagery scripts. Subjects were instructed to image during tape-recorded scripts of standardized neutral (Neutral) and personally relevant fear (Phobic) scripts. All scripts contained both stimulus and response cues; however, subjects were presented four versions of a script which varied in response cue (propositional) content and presentation style: a version presented non-interactively which contained no cognitive cues (meaning propositions), a version presented non-interactively which contained cognitive cues, a version presented interactively which contained no cognitive cues, and a version presented interactively which contained cognitive cues. Both diagnostic groups produced significantly increased HR and SCL in response to Phobic scripts which contained cognitive cues and were presented interactively. Phobic Scripts which contained cognitive cues and were presented non-interactively produced significantly increased arousal only in the panic disorder group as measured by HR. Simple phobics also responded with significantly increased SCL to Phobic scripts presented interactively with no cognitive cues. Neither diagnostic group responded with increased arousal to the Phobic script presented non-interactively with no cognitive cues. Contrary to previous research, these results indicate that subjects with panic disorder with agoraphobia are capable of producing significantly increased physiological arousal in response to phobic imagery. The crucial importance of imagery script content and presentation style are highlighted by the results. Furthermore, the current investigation differed from previous investigations in that the parameters of an actual therapy session were more closely approximated by having the subjects image during script presentation rather than subsequent to script presentation. Finally, frontalis EMG did not prove to be a sensitive measure of anxiety in these subject populations pointing to the need for multiple channels of physiological measurement. Implications for content and methodology of future research studies in this area are discussed.