Reasoning biases such as jumping-to-conclusions (JTC) and incorrigibility have been suggested to contribute to the generation and maintenance of delusions. However, it is still debated whether these biases represent stable traits of patients with delusions, or are related to state fluctuations of delusion severity. The present study aimed to elucidate this question by combining a cross-sectional with a longitudinal approach. JTC, incorrigibility and delusion severity were assessed in 79 patients with a history of delusions over a 6-month period. To allow for a differentiated look into effects of time vs. symptom changes, patients were divided into patients with (D+) and without (D-) current delusions at baseline. Significant improvement of delusions was noted in D+ at follow-up. JTC did not differ between the two patient groups either at baseline or over time. In contrast, incorrigibility was significantly higher in D+ than D- at baseline; this difference remained stable throughout the 6-month follow-up period. The two biases did not significantly co-vary over time. Our results suggest a dissociation between incorrigibility and JTC as regards their relation to current presence of delusions, and tentatively support theoretical accounts attributing different roles to the two biases in the generation (JTC) and maintenance (incorrigibility) of delusions.