The problem of analyzing and evaluating investment risks in agriculture is reviewed in the article. The need to take into account sectoral investment risks when making management decisions on investment location has been underlined. A comparative analysis of the methods and approaches to risk assessment has been carried out, with their main advantages and disadvantages presented. It is proposed to use the method of Kohonen’s self-organizing maps for analyzing sectoral investment risks, taking into account the following indicators: the level of inflation sustainability of the industry’s products, the level of product profitability, the level of competition, the ability to market buyers, the level of state support, the level of social tension. Evaluation of quantitative indicators based on statistical data and quality indicators were assessed by scores. The branches of the agribusiness industry were grouped in clusters corresponding to the level of investment risk based on a neural network constructed using the Deductor Studio Academic program. The result of the analysis determined that the lowest level of investment risk was in investing in growing sunflower, beet, vegetables. Average investment conditions have developed in the following areas: grain farming, pig and dairy industry. Unfavorable conditions for investment have developed in the meat industry and livestock raising. The effectiveness of this method lies in integrating the properties of objects, speed of processing multi-dimensional data and a visual representation of the results. As a result of the study, we constructed a neural network model, which has allowed to group branches by level of investment risk. The neural network presented in this article can be used as a means of information support for decision-making during the determination of the sectoral structure of the investment portfolio.