Data for the study were obtained from one large operational command unit of West Midlands police (England). Data were obtained for all domestic burglaries committed from September 1997 to May 31, 2001, and all data anomalies and errors were corrected. A total of 56 variables were identified for each burglary. The variables were grouped as follows: location of entry, entry methods and behavior, type of dwelling, search behavior, location of exit, alarm/phone, and bogus official crime. Information on revictimization was also collected. A comparison of the number of burglary victimizations for each residence found that most houses were victimized once, and most repeat victims were victimized twice. The study examined whether the modus operandi of the offender distinguished houses burgled only once from those that had more than one victimization. A comparison of burglaries that were "non-repeats" with those that were "quick" repeats (within 365 days) found the distinguishing features to be searching behavior, type of property, entry method, and a bogus official strategy. Survival analyses of "quick" revictimizations found that ramming and removing glass were significantly associated with being revictimized sooner rather than later. In contrast, exit through a window indicated a long period before revictimization. These findings indicate the value of crime-scene information collected by the police in developing burglary-prevention measures that can prevent revictimization. The authors also conclude that this same approach can be used in the analysis of other types of offenses. 5 tables and 33 references