The present study investigated whether a covariation bias is present in severe spider phobics and whether such bias is modified by successful treatment. In addition, this study sought to examine whether a covariation bias is linked to differential autonomic responding. Subjects were 20 untreated phobics, 19 treated phobics, and 18 no-fear controls. Subjects were exposed to a series of 72 slides comprising three categories: spiders (fear-relevant), mushrooms, and flowers. At slide offset one of three possible outcomes occurred: a shock, a tone, or nothing at all. All slide-outcome combinations occurred equally frequent. The results show that an equally strong covariation bias is present in severe spider phobics, in successfully treated phobics, and in no-fear controls. Thus, the present data only partially sustain earlier findings of Tomarken A.J., Mineka S. & Cook M. (1989) (Journal of Abnormal Psychology, 98, 381-394). The covariation bias appeared to be mimicked by differentially heightened autonomic responding. The current data suggest that both the covariation bias and the heightened physiological responding reflect a "beloningness" between spider slides and aversive outcome. © 1991.