By 2030 it has been planning to support more than 30 thousand entrepreneurial and innovative projects of students and graduate students, this idea has become one of the government initiatives in the sphere of social and economic development of Russia. It is known that the pursuit of quantity can lead to a decrease in quality and the subsequent closure of innovative projects. However now there is no generally accepted methodology that allows preliminary assessment of the quality of innovative projects of universities. The presented study empirically illustrates the ap-plication of machine learning methods for assessing the quality of entrepreneurial innovation projects of universities and predicting their success. The study analyzed the main reasons for the closure of Russian startups, resulting in a list of criteria for evaluating the quality of university innovation projects. A survey of managers of innovative projects implemented in universities in St. Petersburg, Moscow and Vladivostok was conducted, on the basis of which a sample was formed for training the quality assessment model of entrepreneurial innovation projects. As a result of the study, a machine learning model was built and trained, and its accuracy was evaluated. Random Forest was chosen as the ma-chine learning method; the reason for its choice was its high accuracy, resistance to outliers, and ability to process data with a large number of features and classes. The application of these results will simplify the process of assessing the quality of entrepreneurial innovation projects and allow predicting their success. Assessing the quality of entrepreneuri-al innovation projects at the early stages of their implementation will be beneficial not only for project managers, but also for business incubators at universities.