Introduction. Anomalies on probabilistic reasoning, theory of mind (ToM) tasks, and attributional biases have been found in delusional people. Delusions are also effectively modified by cognitive behavioural therapy (CBT). We sought to examine whether CBT reduces delusional conviction by changing such general reasoning anomalies. Method. Sixteen patients commenced an 8-11 week CBT programme that targeted their delusions. Probabilistic reasoning, attributional biases, and ToM were assessed pre- and post-treatment. Delusional conviction, preoccupation, and distress were rated at each session. Pretreatment task performances were compared to norms. Repeated measures analyses compared pre- and posttreatment task performances and ratings of delusions. Correlational analyses were used to identify factors associated with reduced delusional conviction. Results. At baseline, 11 patients showed some form of abnormal probabilistic reasoning, 13 excessive attributional biases, and 13 defective ToM compared to norms. Fourteen patients completed the CBT programme and showed significant reductions in delusional conviction and preoccupation. Despite some inconsistent evidence of improvement in verbal ToM tasks, reasoning styles in these 14 patients were largely unchanged by CBT. Conclusion. Reasoning anomalies associated with delusions in this sample mark a vulnerability that persists and is independent of the effectiveness of CBT.