Maladaptive cognitive beliefs as measured by the Dysfunctional Attitudes Scale (DAS) increase vulnerability to depression. Maladaptive metacognitive beliefs as measured by the Metacognitive Questionnaire-30 (MCQ-30) are also thought to contribute to depression. However, the long-term stability of metacognitive beliefs in depression has not yet been investigated. It is unclear whether metacognitive beliefs can add explanatory power to depression above and beyond maladaptive cognitive beliefs. The aim of the present study was to investigate the role and stability of cognitive and metacognitive maladaptive beliefs in depression. Eighty-four patients with depression were assessed with the DAS, three subscales of the MCQ-30 (positive metacognitive beliefs about worry and rumination [PB]; negative metacognitive beliefs about the uncontrollability of rumination [NB]; metacognitive beliefs concerning the need to control one's thoughts [NFC]), the Hamilton Depression Rating Scale, and the Beck Depression Inventory at baseline and were reassessed 3.5 years later. Analyses using a longitudinal latent growth model showed that change on the DAS and baseline scores and change on the MCQ-30 (NB and NFC) significantly predicted change in self-rated depressive symptoms over 3.5 years. However, the DAS explained more additional variance than the integration of the MCQ-30 subscales. Subscales of the MCQ-30 were more stable than the DAS. Although cognitive and metacognitive maladaptive beliefs were both predictors of depression, the DAS was a better predictor than the MCQ-30 subscales. Nevertheless, because maladaptive metacognitive beliefs were more stable than maladaptive cognitive beliefs, they should be considered an important underlying vulnerability factor for depression.