An armed robbery alarm system was implemented in 48 different stores in 2 separate geographical areas in Nashville, Tenn., for 6 months and 12 months, respectively. These 2 target zones were located in a 12-zone area accounting for 85 percent of all Nashville business robberies. The alarms were placed in the two separate areas at different times and all alarms were eventually removed. Thus, multiple baseline and reversal strategies were used to evaluate program impact. A device planted in a cash drawer was triggered whenever 'bait' money was removed from the drawer, sending an alarm signal directly to police cars and headquarters. Onscene apprehensions of armed robberies within target stores were greatly increased, even though the armed robbery systems did not deter robbery incidents nor influence the court disposition of the cases. There was also no crime deterrence, crime displacement, or increased apprehensions in either the immediate neighborhoods of target stores or on a city-wide basis. The cost effectiveness of the program, assessed by analyzing cost per arrest ratio, was calculated to be poor, although the program is being maintained because of the absence of an alternative robbery apprehension technology. Because no crime reduction occurred, the only traceable direct cash benefit was in improved property recovery rates. Ways to improve cost-benefits include the following: (1) clearly identify stores protected by the system, (2) rotate the system among smaller samples of these stores, and (3) place systems in unadvertised stores. Program modification included the addition of camera attachments. Nine references and two tables are provided. (Author abstract modified)