The purpose of this research is to develop a methodology for modeling the probability of a project’s success and to realize a prototype model to develop recommendations for more efficient project and portfolio management in a bank. The work contains 40 foreign and 10 Russian-language sources. The first part of the study covered the theoretical foundations of project and portfolio management. The features of project management in banks are analyzed, the examples of factors and criteria for project success are given. Approaches to modeling in problems of binary classification were considered. In the second part, the main prerequisites, assumptions and limitations were described, a high-level model concept was proposed, and the author’s own modeling methodology was developed for estimating the probability of project success. In the third part, a prototype model for assessing the probability of project success was created. In addition to the tasks performed, recommendations were developed based on the results obtained: factors that predetermined the success of the project were identified and a flowchart of the implementation of the model system was compiled, consisting of a static and dynamic module into the project / project portfolio management processes. Further research directions were identified, including testing other modeling approaches (for example, using neural networks), assessing the economic effect of introducing a system of models, clustering projects in order to identify groups of similar features and apply appropriate management influences to successful project management.