The sample used to test the model consisted of 236 banking offices of State-chartered members of the Federal Reserve System operating in the Third Federal Reserve District. The majority of the observations were of offices in the Philadelphia Standard Metropolitan Statistical Area, with the remainder in rural New Jersey and Pennsylvania. Data on the number of robbery attempts at each office and information on the existence of guards, cameras, and bullet-resistant barriers by office (to be measured by dummy variables) were obtained from bank questionnaires (the period was 1975). The data on police response times were also obtained from bank questionnaires. Results suggest that the type of neighborhood surrounding the bank location and the presence of guards in the banking office influence the risk of bank robbery in the expected directions. Banking offices operating in center-city locations run a relatively high risk of victimization, all else being equal; so, too, do banking offices located in ghetto areas, suggesting that bank robbery may be relatively local in nature, at least in such areas. Guards significantly reduce the risk of robbery. Accepting point estimates, the magnitude of the reduction is about one robbery attempt a year for those offices which would have otherwise suffered a positive number of robbery attempts. These findings accord with Camp's interview of 157 imprisoned bank robbers, most of whom reported ignorance of most of the security data examined, with the exception of knowledge about the presence of a guards. In only 23 percent of the robberies did the robber not know if the bank had a guard. Tabular data and 18 footnotes are provided.