The paper describes an analysis of the relationship between changes in the number of guns and changes in the number of crimes. Both variables are non-negative integers with large mass points at zero, and both variables are likely to affect each other. We account for these characteristics by analyzing our data with a multivariate Poisson-lognormal model that we estimate with the Gibbs sampler. Because county-level data on gun ownership are not available, we use data on subscriptions to the gun magazine Handguns Magazine as a proxy. We do not find any evidence of a pattern that would suggest that either more gun magazine sales lead to more crimes or that more crimes lead to more gun magazine sales.