Coronavirus is a disease that affects the lives of millions of people around the world. Predicting its development is very important because it allows you to plan the load on the health care system, the required number of medical staff, medical and laboratory equipment, the number of free beds in hospitals and the introduction of restrictive measures such as lockdown. The paper considers and applies the classical SIR model for forecasting the epidemic in Ukraine and found that this model in its simplest form is suitable only for long-term forecasting of the general direction of the epidemic, provided there are sufficient reliable statistics on the epidemic, which are currently missing. As an alternative, a statistical approach to forecasting using the Holt-Winters time triple exponential smoothing method is proposed. A website has been created for a dynamic online display of forecasting results, which can be found at this link: https://covid19-info.github.io/covid/index.html. keywords: coronavirus disease, SIR model for predicting the development of epidemics, time series analysis, Holt-Winters method.