The authors develop a causal model of the determinants of carrying firearms for protection. It is argued that the examination of carrying a gun as a dependent variable is methodologically more advantageous than using gun ownership as the dependent variable. Through use of path analysis, the direct and indirect effects on gun carrying behavior are identified. The results suggest most crime specific variables, e.g. fear of criminal victimization, do not directly influence the tendency to carry a gun, although these factors had minor indirect effects on the dependent variable. The most important direct influences on carrying a gun were age, gender, and the belief that the presence of guns is an effective deterrent to crime.