The crop insurance sold in Brazil is estimated using the average of the municipal agricultural productivity of the last four or five years. Many researchers believe that using a probabilistic model for the distribution of agricultural productivity is a better methodology for this purpose. This study advocates using the skew-normal distribution as an alternative to the normal distribution to model the distribution of agricultural productivity in Brazil. We analyzed series of corn yield from 1981 to 2007, in 30 counties in Paraná State, Brazil. The skew-normal distribution showed better fits than the normal distribution for the vast majority of the counties, and, consequently, led to better estimates for the expected payment of crop insurance.