The analysis addressed the historical question of why Detroit robbery killings increased substantially in the mid-1960's, but focused on a test of the hypothesis that the use of guns in robbery elevates the risk of death for the victim. The causal model which guided the data analysis predicted that the number of robbery killings is a function of the number of robberies, the proportion of robberies committed with a gun, patterns of offender intention, and a set of disturbance variables assumed to be independent of the predictor variables. The Detroit data fit a model in which the number of robberies and the proportion of robberies with a gun increase the risk of the victim being killed in the robbery. The basic conclusions remain the same whether the variables are represented as frequencies, incidence rates, or logarithms of incidence rates. Results also persist in the presence of controls for demographic changes. 7 notes, 15 references, 5 tables.