Abstract. Cognitive-behavioral therapy for psychosis (CBTp) builds on theoretical models that postulate reasoning biases and negative self-schemas to be involved in the formation and maintenance of delusions. However, it is unclear whether CBTp induces change in delusions by improving these proposed causal mechanisms. This study reports on a mediation analysis of a CBTp effectiveness trial in which delusions were a secondary outcome. Patients with psychosis were randomized to individualized CBTp (n = 36) or a waiting list condition (WL; n = 34). Reasoning biases (jumping to conclusions, theory of mind, attribution biases) and self-schemas (implicit and explicit self-esteem; self-schemas related to different domains) were assessed pre- and post-therapy/WL. The results reveal an intervention effect on two of four measures of delusions and on implicit self-esteem. Nevertheless, the intervention effect on delusions was not mediated by implicit self-esteem. Changes in explicit self-schemas and reasoning biases did also not mediate the intervention effects on delusions. More focused interventions may be required to produce change in reasoning and self-schemas that have the potential to carry over to delusions.