Several studies have indicated that phobic fear is accompanied by a covariation bias, i.e. that phobic Ss tend to overassociate fear relevant stimuli and aversive outcomes. Such a covariation bias seems to be a fairly direct and powerful way to confirm danger expectations and enhance fear. Therefore, it has been suggested that covariation bias is an important factor in the maintenance of phobic fear. However, thus far there are no empirical data available to exclude the alternative possibility that covariation bias is a mere epiphenomenon of fear. To explore the "causal" status of covariation bias, successfully treated spider phobics who participated in an earlier study on covariation bias were asked to complete a Spider Phobia Questionnaire at 2 yr follow up. Results indicate that Ss who displayed a covariation bias immediately after treatment are more vulnerable to relapse than Ss who did not show such a bias. This finding strengthens the idea that covariation bias may enhance fear, thereby contradicting the suggestion that covariation bias is a mere epiphenomenon of phobic fear.