Problem statement. The task of the research is evaluation the dynamics of changes in the level of injuries at work depending on the main causes of the accident. Technical, organizational, psychophysiological, technogenic, social, natural and environmental causes are closely interrelated and have a significant impact on the number of victims of accidents at the industrial plants, as well as on the number of victims of accidents with fatal consequences. The purpose of the article. The assessment of changes in industrial injuries at Ukrainian enterprises during 2010−2022. The establishing connection between social and industrial factors affecting the rate of industrial accidents. The creation of correlation-regression models for statistical evaluation and analysis of the influence of factor variables on the results of injury. Methodology. The use of descriptive statistics for the analysis of the dynamics of changes in the level of industrial injuries. Carrying out a correlation analysis to establish the density of connection between factor variables and resulting features. The use of regression-variance analysis to obtain coefficients of regression mathematical models and statistical indicators that explain the probability of significance of these coefficients. Calculation of the value of the relative error of the calculation data obtained according to mathematical models to confirm their adequacy. Scientific novelty. Multiple correlation-regression models have been developed that take into account the main causes of the accident as factor variables affecting the injury rate and the rate of fatal accidents. Practical significance. Mathematical models make estimate the impact of the main causes of accidents on the level of industrial injuries. The implementation of improved methods and means of safe activity at work is an important tool for correcting the cause-and-effect relationships of industrial injuries. Conclusions. The correlation-regression models were created to analyze the level of industrial injuries in Ukraine. The numerical calculations were carried out according to these models. The average values of the relative errors of the calculated data are 1.55 % and 6.08 %, which indicates the adequacy of the developed models.