In December-2019 China reported several cases of a novel coronavirus later called COVID-19. In this work, we will use a probabilistic method for approximating the true daily numbers of infected. Based on two distribution functions to describe the spontaneous recovered cases on the one hand and the detected cases on the other hand. The impact of the underlying variables of these functions is discussed. The detected rate is predicted to be between 5.3% and 10,8%, which means that there would be about 38 million infected until now (10-May 2020), rather than the officially declared number of 3.99 million worldwide cases.