We show that although Gibbs and Erickson fail to introduce an important control (metropolitan population), the relationship persists after such a control is introduced. We also show that the relationship holds for a much wider range of cities than the 32 large cities studied by Skogan. Differing explanations of the relationship offered by Gibbs and Erickson and by Skogan are evaluated. The Gibbs and Erickson explanation involves implausible assumptions about movement of crime participants from suburbs to cities. The relationship is more likely explained on the basis of urban-suburban status differentials and central-city decline, both of which are maximized in larger, older, more suburbanized cities. (Author abstract)