This research has been done with the aim of evaluating bank non-performing loans from the point of view of the level of difficulty in collecting the debts related to them. In the first part, which includes the preparation of a tree structure including the factors and characteristics that determine the difficulty of NPL cases, library studies, in-depth interviews, semi-structured interviews and Delphi sessions have been used, and in the second part, the purpose of which is to determine the effectiveness and desirability of the mentioned factors and characteristics. AHP and BWM methods have been used to analyze the opinions of 50 experts in the field of credit and collection of bank debts. Expert choice and Python software were used to perform the analysis. Three factors of facility and debtor's background, collection tools, and debtor's non-current debt status were identified as the main effective factors in the difficulty of debt recovery and were divided into 40 effective sub-factors in a tree structure. The possibility of seizing, acquiring and selling the client's immovable property was determined to be the most important factor in the evaluation of the case. With the model obtained in this research, it is possible to evaluate all the bank claims files and grade their difficulty levels.