Results indicated that citizens and the police responded differently to crime prevention activities in different neighborhoods. Citizens living in high-risk neighborhoods were more likely to be involved with community crime prevention efforts than citizens in low- to moderate-risk neighborhoods. The strongest predictor of involvement in crime prevention in all types of neighborhoods was one measure of social cohesion--whether citizens felt like they were part of the neighborhood. The other measure of social cohesion--whether citizens relied on neighbors for help--was not a significant predictor of crime prevention involvement. Other findings revealed that minority citizens were more likely to be involved in crime prevention planning than Whites. The findings suggest that citizen involvement in high-risk neighborhoods is impacted by a combination of personal, parochial, and public social control mechanisms operating in these communities. The authors suggest that strategies to improve private networks of informal social control should be developed and implemented in high-risk neighborhoods. Data were drawn from a 1997 random telephone survey of 2,362 Boston residents on public safety issues. Incident reports for violent crimes (homicide, rape, assault, and robbery) and all arrest reports for 1997 were gathered from the Boston Police Department. Measures of citizen involvement in crime prevention activities in their neighborhood during the previous year were based on self-reports. Data analysis involved the use of multivariate logistic regression models to compare predictors of involvement in crime prevention across neighborhoods with varying levels of crime. Figure, table, notes, references