Alterations in decision-making have been suggested to underlie different psychiatric illnesses, such as depressive disorder (MDD), obsessive-compulsive disorder (OCD), and schizophrenia. Studies report overlapping and divergent alterations in individuals diagnosed with these disorders linking to their respective symptoms. As these disorders often occur as comorbidities, it is crucial to disentangle alterations in decision-making that are unique to specific disorders, from those that occur across diagnostic boundaries. Recent developments in computational modelling have enabled a mechanistic deconstruction of decision-making processes, facilitating the identification of disease-specific and disease-general alterations. We, therefore, used two computational modelling approaches to a two-step Markov decision task in a transdiagnostic cohort combining individuals with MDD (N=23), OCD (N=25), and schizophrenia (N=27) to identify disease-specific and disease-general decision-making alterations, and associations with symptoms. The results revealed overall reliance on model-free decision-making behavior across all groups. Additionally, schizophrenia patients showed lowest learning rate in stage one and highest switching rate (lowest perseverance) compared to all other groups. All patient groups were more random than controls in stage two, with schizophrenia patients revealing highest levels of randomness, while OCD patients were significantly more random in stage one compared to controls and MDD patients. Importantly, the findings revealed a correlation between increased model-free behavior and elevated depressive symptoms, while more model-based choices were linked to lower levels of anhedonia across all patients. Consequently, the findings emphasize noteworthy disease-general and disease-specific decision-making alterations in MDD, OCD, and schizophrenia patients, suggesting that concurrent alterations and symptoms might potentially have common underlying behavioral mechanisms.