This study examined the influence of the southern brand of conservative Protestantism in sustaining and legitimating distinctive forms of violence. To demonstrate the existence of a subculture of violence, at least two conditions must be met: (1) members of the hypothesized subculture exhibit distinctive norms and practices with regard to the appropriate or inappropriate use of violence, and (2) there are identifiable mechanisms of socialization via which these norms are legitimated, sustained, and transmitted to new members. One hypothesis of this study was that southern metropolitan statistical areas (MSAs) would have higher homicide rates than those in other regions of the United States, even with controls for structural variables. The second hypothesis was that the proportion of MSA residents affiliated with conservative Protestant faiths would be positively associated with homicide rates. The third hypothesis was that controls for the proportion of conservative Protestant residents would substantially reduce or eliminate the association between Southern region and homicide rates among MSAs. The fourth hypothesis was that the estimated net effects of conservative Protestantism would be stronger among Southern MSAs, as compared with MSAs in other regions. The study used 1980 data on homicide rates, religious concentrations, and socioeconomic and other characteristics of 247 MSAs. Several aspects of Southern religious culture are identified, such as tendencies toward dichotomous morality, and dispositional attribution styles. After controlling for a number of relevant covariates, this study found a positive relationship between percent conservative Protestant and homicide rates; the effect is significant in Southern MSAs but not in non-Southern MSAs. Religion may be only one component of a cultural explanation of Southern violence. Further research is needed to explore the spatial and temporal variance of these regional and religious effects. The possible influence of regional religious culture should be explored in studies of other outcomes. 4 tables, 2 appendices, 64 references