The Poisson model, discovered nearly two centuries ago, is the basis for analyses of rare events. Its first applications included descriptions of deaths from mule kicks. More than half a century ago the Poisson model began being used in geographical analysis. Its initial descriptions of geographic distributions of points, disease maps, and spatial flows were accompanied by an assumption of independence. Today this unrealistic assumption is replaced by one allowing for the presence of spatial autocorrelation in georeferenced counts. Contemporary statistical theory has led to the creation of powerful Poisson-based modeling tools for geographically distributed count data.