The authors investigated the role of phobic responsivity in the generation of phobia-relevant illusory correlations. As a means of disentangling the contributions of prior fear and elicited fear responses, half of a group of phobic women received 1 mg alprazolam (n = 21), and half received a placebo (n = 22). A group of nonfearful women (n = 24) was included to control for prior fear per se. Participants were exposed to slides of spiders, weapons, and flowers that were randomly paired with a shock, a siren. or nothing. Postexperimental covariation estimates and on-line outcome expectancies were assessed. Irrespective of both prior and elicited fear, participants postexperimentally overassociated spiders and shock. Yet, only women with spider phobia displayed a persisting fear-confirming expectancy bias. This bias was similar for the placebo and alprazolam groups. Thus. the bias appeared to be due to preexisting phobogenic beliefs, whereas phobic responsivity played a negligible role.