Fulfillment of construction projects on time in energy and other sectors of the Russian economy is investigated. The losses caused by the failure to complete projects on time are noted to be significant. The article discusses the risks negatively affecting project management and analyzes the approaches to project management considering the risk of failure to complete projects on time. Based on the analysis of RusHydro’s activities in recent years, a map of the risks of the company has been drawn up. It is clear from the map that the risk of failure of completion dates is the most critical for the company. The risk of higher cost of work is just as critical and often associated with the risk of delay, as it is a consequence of the implementation of measures to reduce the latter. We analyzed the approaches to project management based on the risk of their non-fulfilment on time. The object of the study is projects to create (design, construct or reconstruct) large energy facilities, the subject is tools to support decision-making in the management of projects in the face of uncertainty and risk. The aim of the study is to improve project management practices considering uncertainties and risks based on the assessment of the economic impact of the decisions made, allowing us to select related management decisions from a variety of alternatives at the successive stages of project implementation. It is proposed to manage the risks of the project during its planning and analysis phase taking into account the alternative of solutions and the stage of the project implementation to combine the PERT-time project management method with one of the methods of risk management – the decision tree method. The general example of applying the proposed combined method of project risk management is considered. The practical significance of the research is noted. The complexities of the method and areas of further research are indicated. Building a solution tree is usually used to analyze the risks of the projects that have a foreseeable number of options. Otherwise, the tree has a very large volume, so it is difficult not only to find the optimal solution, but also to determine the raw data.