The aim of the study was to develop a step-by-step system for predicting late spontaneous miscarriages, taking into account clinical, anamnestic, echographic data, and the cytokine status of pregnant women. Materials and methods. The comparative study consisted of two stages. A total of 260 women were included in the study. At the retrospective stage, the main group consisted of 66 patients with late spontaneous miscarriage at a gestational age of 10.0-21.6 weeks (17.8 ± 2.1 weeks). The control group included 90 patients whose pregnancy ended in term delivery. The prospective stage included patients with threatened late spontaneous miscarriage at a gestational age of 10.0-21.6 weeks (main group, n = 51), who were divided into two groups depending on the pregnancy outcome. The clinical and anamnestic, echographic data were assessed, the concentration of IL-6, IL-8, IL-10 in the blood serum was determined by the enzyme immunoassay method using reagent kits. The data were processed using discriminant analysis, logistic regression and elements of mathematical modeling. Results and discussion. The results obtained in the study allowed us to create a step-by-step forecasting system. The first developed prognostic model of the risk of late spontaneous miscarriage, based on clinical and anamnestic data, allows us to predict the risk of late spontaneous miscarriage with high efficiency, while 99.5% of cases are correctly classified. This model is not quite effective (AUC 0.60) in terms of predicting the implementation of a threatening late spontaneous miscarriage, for this purpose, the content of interleukins-6, -8, -10 in the blood was studied as potential predictors. Based on the obtained results, a prognostic model 2 was created, which allows assessing the risk of implementation (unfavorable outcome) of a threatened late spontaneous miscarriage. Conclusion. The proposed algorithm of step-by-step forecasting will allow for correct stratification of the risk of late spontaneous miscarriage and personalized pregnancy management to reduce the risk of unfavorable outcomes.