An increasing number of cases of infection and death by COVID-19 has been observed in several parts of the world, including Brazil. While scientists are looking for a drug / vaccine capable of combating COVID-19, its devastating action is spreading out of control. In this context, statistical studies and preliminary analyzes of the epidemic situation may be important to provide a basis for disease prevention and control. Thus, the objective of this work was to adjust nonlinear regression models to mortality data and confirmed cases of COVID-19 in Brazil, Italy and the world until 03/31/2020. Data from the Ministry of Health of Brazil and the World Health Organization were used. The models were compared using the Akaike information criterion and the Bayesian information criterion, as well as the determination and adjusted determination coefficients, in addition to the square root of the mean square error. All models presented were adequate to model the studied variables. It is not yet possible to make reliable projections of when the numbers of confirmed cases and deaths will decrease. Social detachment in Brazil is being effective in restricting the progression of the disease by reducing the speed of infection and transmissibility.