Abstract: Nowadays stress testing is becoming an important role as a mandatory tool of risk management in the modern banking activity. One of the most important stages of stress testing is the construction of shock scenarios, which on the one hand would have a historical basis, and on the other — take into account the depth of a possible crisis. The relevance of the research topic is due to the fact that on the one hand the use of models in stress-test scenarios is a mandatory regulatory requirement, and on the other hand that it is incorrect to create a universal method of stress scenarios, because shock scenarios developed on an individual basis is more effective due to taking into account the strategy and business model of each bank. When conducting macroeconomic research in the field of stress testing, we constantly have to face the need to assess the impact of various shocks on the economic dynamics in the country. In addition, many time series, especially in the financial sector, change synchronously in a certain way. Therefore, the development of an econometric model, which will describe the interaction between different financial time series in unstable economic conditions, plays an important role in the process of stress testing of market risk. The authors chose the vector autoregression model as such a model. New scientific result of the publication is the development of reasonable scenarios of stress testing of market risk based on the methods of vector-autoregressive analysis. The authors’ methodology of construction of stress scenarios of market risk with the use of vector autoregression models is offered in the article. As endogenous variables of the econometric model, we use the main risk factors of market risk, which reflect currency and interest rate risks: official exchange rates of dollar and euro against hryvna, short-term and long-term interest rates of loans in hryvna, dollar and euro. The calculations were performed in the software package EViews 10.0. The test did not reveal cointegration between variables, so it was evaluated the standard VAR model with constraints. Impulse response functions were constructed and analyzed, which are considered as stress scenarios for assessing market risk. The results of scientific research can be important to risk units of commercial banks that implement tools of market risk stress testing in the internal risk management system.