Crime is one of the most important social problems for administrative region. Ascertaining the detailed characteristics of crime and preparing countermeasures are important to keep community life safe and secure. A lot of studies using crime data and geographical data have been carried out with a view to crime prevention. These studies include analyzing geographical features of crime, mapping crime-related information and crime hotspots on the map, predicting crime rate and so on. In addition, police stations have recently begun emailing notifications regarding crime to citizens to help them avoid crime. The e-mail messages include rich information about regional crime; they are actively used by services providing guidance to people in how to avoid crime. These services map the messages onto regional maps using the location information in the messages and show the relations between the locations and crime on the map. In addition, some services send alarms to their users when the GPS information of the users indicates that they are passing by the places where crime has occurred. However, these services only use the location and crime information extracted from the messages. Thus, we cannot say the messages have been fully used to clarify characteristics of regional crime. Therefore, in this paper, we investigate whether or not the crime messages sent by e-mail can be further exploited as a valid source for analyzing the criminal characteristics of a region, i.e., whether or not they include the characteristics of regional crime. To this end, in this research, we conducted experiments to make clear whether or not the crime messages sent by e-mail can help to distinguish regions. Experimental results illustrate that the contents of e-mail crime messages helped to distinguish regions having greater than or equal to 100 reports, with an average F-measure of about 90.3%, while only using the names of the areas where crime has occurred cannot match that F-measure.