Subway riders were selected for interviews by using a quota sample of 2,700 passengers on trains serving a 24-station area. These stations were chosen because they were perceived as being high-crime areas of the subway system. Respondents were selected from passengers riding trains as well as those waiting for trains in subway stations. Subjects were queried about the carrying of weapons on their person, their fear of crime, and past victimization. Interview findings indicate that at least one-fourth of the subjects carry some form of self-protection, ranging from guns and knives to mace. A number of factors are associated with inclinations to carry self-protective devices. Among them are fear of crime and of victimization; previous victimization; and the gender, age, and race of the respondent. The strongest predictors and those that contribute to the best multivariate model are previous victimization and fear of victimization. Police generally believe that citizens' resort to self-protective weapons is an ineffective and potentially dangerous strategy for crime prevention. This behavior is only likely to diminish, however, if alternative, more effective, and safer crime prevention measures are devised such as changes in the physical environment and the use of citizen groups in appropriate ways to deter crime and reduce the fear of crime. 2 tables, 4 notes, and a 17-item bibliography